Anis Yazidi

• Sengupta, Poushali; Khadka, Rabindra; Maharjan, Sabita; Eliassen, Frank; Pandey, Shashi Raj & Zhang, Yan [Show all 8 contributors for this article] (2026). Explainability of Complex AI Models with Correlation Impact Ratio. IEEE Transactions on Artificial Intelligence (TAI). p. 14–14. doi: 10.1109/tai.2026.3732283. Full text in Research Archive
• Bhandari, Shailendra; Szorkovszky, Alex; Yazidi, Anis & Lind, Pedro (2026). Evolutionary foraging in grids: Intermittent search dynamics emerge in finite, depletable landscapes. arXiv. doi: 10.48550/ARXIV.2609.39239. Full text in Research Archive Show summary

  How search strategies evolve in finite, depletable landscapes remains a central question in foraging theory. We study this problem with an evolutionary simulation in which agents forage on a two-dimensional toroidal lattice containing non-renewable resources distributed either uniformly or as L\'evy dust. Each agent carries a heritable genome encoding step lengths, velocities, and turning angles, and selection acts on a fitness function combining energetic gain, movement cost, and coverage efficiency. By allowing movement traits to evolve without imposing a prescribed power-law step-length distribution, we test whether the evolved trajectories are better described by intermittent-search or L\'evy-walk dynamics. Strikingly, our results indicate that, in the finite depletion-driven landscapes considered here, evolved search is more consistent with intermittent dynamics than with strict scale-free L\'evy motion. To characterize the effective dynamics of the evolved trajectories, we take the average adjusted coefficient of determination when fitting second- and fourth-order displacement moments to intermittent search and L\'evy-walk models. While a L\'evy-like random walk fits with the numerical results from the evolutionary search very well (mean adjusted \(\bar{R}^2>0.9\) in most tested conditions), the intermittent search achieves a closer fit, with mean adjusted coefficients of determination(\(\bar{R}^2>0.99\)) for all tested resource distributions. This preference holds across the tested grid sizes and resource densities. Five independent evolutionary runs per environment on a $503\times503$ grid at nominal resource density $\rho=0.15$ reproduce this preference for the uniform environment and the five L\'evy-dust environments. Evolution rapidly reshapes the movement genome toward short displacements while retaining a sparse tail of longer relocations, consistent with local exploitation punctuated by occasional transfer. The framework provides a controlled setting for studying how search rules emerge under resource limitation and may inform resource-constrained exploration in autonomous systems. Project webpage: \url{https://evo-foraging.github.io/}

• Basnet, Laxman; Szorkovszky, Alex; Lind, Pedro; Yazidi, Anis & Bhandari, Shailendra (2026). Diffusion models for eye-gaze trajectory generation using position and velocity representations. arXiv. doi: 10.48550/ARXIV.2609.05522. Full text in Research Archive Show summary

  Eye-tracking data are expensive to collect, requiring specialized hardware and controlled laboratory conditions, and difficult to share due to privacy regulations, which together limit the quantity and diversity of gaze data available for research. We address this with two complementary denoising diffusion probabilistic models (DDPMs) for unconditional generation of continuous eye-gaze dynamics from the same visual-search eye-tracking dataset. Both use an identical FiLM-conditioned one-dimensional U-Net with self-attention (19.35\,M parameters) trained on data from visual searches performed by 28 participants, with sliding-window samples 8\,s in length. The first model generates raw two-dimensional gaze-position sequences directly, and the second generates the corresponding two-component velocity sequences; the two formulations use representation-specific preprocessing, training configurations, data partitions, and evaluation protocols. Each model is evaluated across three independent training seeds over its own fixed train/validation split; seed-aggregated metrics are reported as mean\,$\pm$\,SD while single-reference-run analyses are identified separately. The position-space model achieves a mean Jensen-Shannon (JS) divergence of $0.016\pm0.004$ over nine kinematic features, with the highest feature-wise mean below $0.030$, fixation duration within 2\% of real data, and a Fr\'echet Gaze Distance more than an order of magnitude below statistical and Markovian baselines. Under a Train-on-Synthetic-Test-on-Real protocol, synthetic-only training achieves $R^2=0.66\pm0.02$, corresponding to 82.7\% of the real-data $R^2$ point estimate. The velocity-space model achieves a mean JS divergence of $0.0065$ across velocity-component, speed, log-speed, and turning angle distributions, with a maximum of $0.015\pm0.005$. Reconstructed path length is less accurately reproduced ($0.21\pm0.02$ versus $0.03\pm0.01$ in position space), although the validation protocols differ. Taken together, the two representations indicate that unconditional diffusion reliably captures the local kinematic structure of human gaze motion, velocity-component marginals, scalar speed, and short-range temporal and directional statistics, while integrated, long-range structure such as saccade counts and cumulative path geometry remains the principal target for future stimulus- or participant-conditioned models.

• Acharya, Sushil; Fabian, Karl; Yazidi, Anis & Westeng, Kjetil (2026). Automated depth alignment of well logs using unsupervised machine learning and change-point detection. Results in Engineering (RINENG). 32, p. 1–17. doi: 10.1016/j.rineng.2026.112513. Full text in Research Archive Show summary

  Accurate depth alignment between Logging While Drilling (LWD) and Electric Wireline Logging (EWL) data remains a time-consuming step in well-log interpretation. This study presents an unsupervised workflow for EWLLWD depth alignment that combines convolutional autoencoders, latent-space clustering, structural boundary matching, and kernel change-point detection, without labeled training pairs or manual tie points. A convolutional autoencoder (CAE) compresses six log channels into a two-dimensional latent space. Joint clustering identifies structurally consistent intervals, whose transitions define candidate matching boundaries. These boundaries are matched under monotonicity, depth, and similarity constraints, followed by local refinement using kernel changepoint detection. The method is evaluated on three wells using 256-sample slices. Pearson correlation and mean absolute error are used as measures of log similarity and structural consistency, and compared with cross-correlation, manual depth alignment, and a re-evaluated supervised CNN benchmark. For well 16/1-9, the proposed method increases the mean slice-wise Pearson correlation for gamma ray (GR) and deep resistivity (RDEP) from 0.76 to 0.86, and for neutron porosity (TNPL) from 0.61 to 0.74. However, improvements vary across log types in magnitude and statistical significance, and some logs, such as PEF, show limited improvement. Across wells 31/3-Q-21 and 6407/10-5, GR-based cross-well evaluation shows that the average GR correlation increases from 0.50 to 0.94 and from 0.30 to 0.88, respectively. These results support the proposed workflow as an unsupervised approach for improving log similarity and structural consistency between EWL and LWD logs, although performance is log dependent and evaluation relies on similarity metrics rather than independent depth-control measurements.

• Gorosito, Martin; Yazidi, Anis; Hermansen, Åsmund; Berg, Bjørnar; Øiestad, Britt Elin & Grotle, Margreth [Show all 8 contributors for this article] (2026). Exploring risk factors for long-term sickness absence during emerging adulthood: Continuous and discrete time models using Young-HUNT data on psychological distress and chronic pain. International Journal of Medical Informatics. ISSN 1386-5056. 219. doi: 10.1016/j.ijmedinf.2026.106551. Full text in Research Archive Show summary

  Introduction: Long-term sickness absence (LTSA) in young adults has important consequences for labour market participation and future work disability. Chronic pain and psychological distress are key risk factors and frequently co-occur, yet their combined impact during adolescence on later LTSA remains insufficiently understood. This study aims to explore factors that influence adolescents’ and young people’s risk of receiving LTSA benefits during emerging adulthood. Methods: This longitudinal study used data from the Young-HUNT1 (1995–1997; n = 8736) and Young-HUNT3 (2006–2008; n = 7935) cohorts linked to Norwegian registry data and followed into early adulthood. The outcome was time to LTSA (90 or 180 days). Associations were examined using Cox proportional hazards models and Kaplan–Meier analyses. Continuous- and discrete-time models were developed and evaluated using the concordance index, time-dependent AUC, and integrated Brier score. Risk factors were analysed using SurvSHAP, SHAP, and regression-based methods. Results: Chronic pain and co-occurring pain and psychological distress were consistently associated with increased LTSA risk (adjusted HRs between 1.3 and 1.5 for pain and between 1.6 and 1.7 for co-occurrence). In contrast, psychological distress alone showed no consistent association. Model performance was moderate and similar across approaches (C-index between 0.63 and 0.67). Key predictors included female sex, low parental education, chronic pain, poor perceived health, and indicators of early health problems. Conclusion: Adolescent chronic pain, particularly when co-occurring with psychological distress, is an important predictor of LTSA in early adulthood. While absolute LTSA levels may vary across cohorts, underlying risk patterns remain stable. More complex models did not outperform traditional approaches. These findings highlight the importance of early-life conditions and support early identification and intervention to reduce later work absence.

• Szczepanski, Tomasz Szymon; Nilsen, Kristian Bernhard; Dunker, Øystein; Reiner, Andrew Henry; Sand, Trond Halfdan & Tannemaat, Martijn [Show all 11 contributors for this article] (2026). Improving Indirect Methods for Calculating Reference Limits for Nerve Conduction Studies From Historical Data. Muscle and Nerve. ISSN 0148-639X. 74(2), p. 328–338. doi: 10.1002/mus.70281. Full text in Research Archive Show summary

  Introduction/Aims: High quality reference limits for nerve conduction studies (NCS) are essential for diagnosis of neuromuscular disorders. Examining healthy controls to calculate reference limits directly is expensive and time consuming. Indirect methods, including extrapolated norms (E-norms), extrapolated reference values (E-Ref), and multivariate extrapolated reference values (MeRef), use historical hospital data instead. These methods sort the historical measurements in increasing order, creating so called S-curves, and select normal measurements based on the shape of these S-curves. Current versions of these methods have several limitations, and we aimed to improve them by modifying how they select normal measurements. Methods: E-norms, E-Ref and MeRef were modified with new S-curve selection algorithms. The modified versions of the methods were used to calculate reference limits for common NCS measurements from a large historical database. The results were compared to reference limits calculated from 680 healthy subjects using Youden’s J statistic and z-score deviation. Results: The modified methods provided reference limits with Youden’s J > 0.8 and z-score deviation < 0.9 for most types of NCS measurements and similar or higher Youden’s J than the unmodified methods. In most cases, the methods required at least 500 measurements and fewer than 20% abnormal measurements for good performance. Discussion: Changing the S-curve selection algorithms improves E-norms, E-Ref, and MeRef. The modifications require a sufficient number and proportion of normal measurements in the historical database. When these prerequisites are met, a combination of indirect methods can be used when developing reference limits from historical patient data.

• Zhou, Hao; Xu, Wentao; Zhang, Xuan; Yazidi, Anis & Yu, Ke (2026). Temporal Conditional Score Network for Multivariate Time Series Anomaly Detection. IEEE Internet of Things Journal. ISSN 2327-4662. 13(13), p. 28165–28180. doi: 10.1109/JIOT.2026.3681780. Full text in Research Archive Show summary

  Time series anomaly detection is a critical technique for secure monitoring in cyber-physical systems (CPSs). Most existing methods learn a normal data profile and identify anomalies as deviations from this profile. Recently, score-based models, which estimate the gradient of the log data probability density, have shown promise in improving data distribution estimation due to their strong generative capabilities. However, their application to time series anomaly detection in CPSs remains underexplored, with key challenges in network architecture design and the accuracy-efficiency trade-off yet to be addressed. In this paper, we propose the Temporal Conditional Score Network (TCSN), a novel framework that jointly estimates conditional scores and detects anomalies based on reconstruction error. Our approach enhances a U-Net backbone with temporal attention skip connections, preserving detailed spatio-temporal features at both variable and time-step levels. Additionally, signature matrix embedding is introduced to improve sensitivity to anomalous patterns. We further establish connections between score-based and reconstruction-based anomaly measurements, leading to a more efficient and accurate detection strategy. To ensure practical applicability, we design a full-cycle deployment pipeline that integrates data sanitization, denoising score matching–based training, hyperparameter tuning, and proactive model updates, enhancing scalability and long-term stability. Extensive experiments on public datasets demonstrate that TCSN consistently outperforms state-of-the-art baselines, achieving up to 0.2155 improvement in composition F1 score while maintaining strong robustness under varying noise levels.

• Acharya, Sushil; Fabian, Karl; Yazidi, Anis & Westeng, Kjetil (2026). Semi-automated bed boundary detection and log alignment using fast Walsh–Hadamard transform. Frontiers in Earth Science. 14. doi: 10.3389/feart.2026.1736164. Full text in Research Archive Show summary

  Depth mismatches between well logs acquired in different runs (e.g., electrical wireline logging (EWL) versus logging while drilling (LWD)) impede stratigraphic correlation and bias reservoir property estimation. Existing alignment methods, such as maximum cross-correlation (MaxCC), are fast but generally assume near-uniform shifts and can be noise-sensitive, whereas dynamic time warping (DTW) can accommodate nonlinear discrepancies but is computationally heavier than global-shift methods and less interpretable. We present a semi-automated Fast Walsh–Hadamard Transform (FWHT) workflow that enhances step-like responses associated with lithological transitions, quantifies boundary evidence via a Walsh Boundary Index (WBI), and uses matched boundary pairs as control points to construct a monotone, piecewise-linear depth-warping function that maps LWD depths onto an EWL reference.“Semi-automated” denotes that, once the log channels and the user-specified cutoff C and WBI threshold τ are defined, all remaining steps are automatic and deterministic. The method is applied to normalized EWL–LWD datasets containing gamma ray (GR), bulk density (RHOB), neutron porosity (NPHI), and compressional sonic slowness (DTC) logs. FWHT increases the slice-averaged Pearson correlation r (mean across logs) from 0.079 to 0.574 and reduces the Euclidean distance e from 1.863 to 0.964; the MaxCC baseline achieves a similar mean correlation (0.595) but a higher distance (1.008), consistent with its global-shift assumption. A constrained global DTW baseline achieves higher similarity ( r = 0.878; e = 0.632), but does so via dense pointwise warping that is less interpretable and can introduce locally nonuniform distortions. Scalability is evaluated on 89 GR log pairs from the Norwegian Continental Shelf: mean Pearson correlation r increases from 0.367 (unaligned) to 0.851 with FWHT (excluding 5 NA pairs), compared with 0.860 using MaxCC. With a uniform threshold ( τ = 0.08 ) , MaxCC performs better in 45 cases, FWHT in 29, and 15 are comparable. Across the evaluated wells, FWHT shows the largest performance gains relative to global-shift methods in intervals exhibiting depth-dependent (nonlinear) misalignment. Overall, the FWHT framework provides an interpretable and computationally efficient approach for boundary-controlled, nonlinear well-log depth alignment.

• Acharya, Sushil; Fabian, Karl; Yazidi, Anis & Westeng, Kjetil (2026). Attention-Driven Convolutional Neural Network for Automated Well Log Depth Alignment. Results in Engineering (RINENG). 30, p. 109901–109901. doi: 10.1016/j.rineng.2026.109901. Full text in Research Archive Show summary

  Well logs often exhibit depth discrepancies between electrical-wireline logging (EWL) and logging-while-drilling (LWD) recordings due to systematic errors, such as cable stretch in EWL and variations in drilling conditions in LWD. These discrepancies can lead to misinterpretations during reservoir analysis. While manual alignment via maximization of cross-correlation (MaxCC) is common, this study proposes an automated Attention-Driven Convolutional Neural Network (AD-CNN). The hybrid architecture incorporates a 1D convolutional neural network (CNN) to extract local spatial features, such as log curve shapes, along with a multi-head cross-attention mechanism that prioritizes critical depth intervals. This design is intended to improve robustness to variable signal quality and depth offsets within the evaluated range. The model is trained on 934,561 standardized gamma-ray (GR) log pairs (151-sample slices) derived from three wells in the Norwegian North Sea, validated on one well, and tested on an independent well. Evaluation also includes four unseen case-study wells, with depth-matching assessed using Pearson correlation, Euclidean distance, and qualitative review of alignment and attention heatmaps. The proposed AD-CNN achieves an average final Pearson correlation of 0.85 across four wells, improving over CNN and MaxCC, and approaching the accuracy of dynamic time warping (DTW) (Pearson correlation = 0.93) while running faster (0.006 s vs. 0.051s per slice). In case studies across four wells spanning different fields, lithologies, and inclinations (up to 36.7°), AD-CNN improves alignment in over 98% of slices, suggesting stable performance across the evaluated conditions. This automated and scalable approach can reduce manual interpretation and support scalable reservoir analysis workflows.

• Mathema, Rujeena; Nav, Shamimeh Mohajeri; Bhandari, Shailendra; Regmi, Manoj; Lind, Pedro & Yazidi, Anis [Show all 7 contributors for this article] (2026). Comprehensive dataset of features describing eye-gaze dynamics across multiple tasks. Scientific Data. doi: 10.1038/s41597-026-06754-x. Full text in Research Archive Show summary

  We present a comprehensive dataset collected from 251 participants, which includes diverse features characterizing eye-gaze dynamics. Participants performed several tasks per session, typical in eye-tracking research: vanishing saccade, cued saccade, flickering cross, rotating ball, and free viewing. Covering these experimental paradigms, the dataset enables analysis of human eye phenomena such as oculomotor control and perceptual processing. As features, the dataset includes timestamped gaze coordinates, pupil sizes, and event classifications, labelled as fixations, saccades, or blinks. All the data were recorded using the EyeLink Portable Duo eye-tracker hardware at 1000 Hz, and processed from raw EyeLink Data Format (EDF) files into structured files via an automated pipeline. The dataset was anonymized according to the Norwegian Agency for Shared Services in Education and Research (SIKT) ethical standards. Possible applications of these datasets are the study of visual attention, cognitive science, and assistive technology.

• Kasahara, Yuri; Schroeder, Daniel Thilo; Yazidi, Anis & Lind, Pedro G. (2026). The Dynamics of Hate Speech: Assessing Anti-Muslim Hate Speech in Norwegian Social Media. Social science computer review. ISSN 0894-4393. doi: 10.1177/08944393261417730. Full text in Research Archive Show summary

  This study investigates the dynamics of anti-Muslim hate speech within Norwegian social media during the period between 2010 and 2021. Using a dataset of more than one million comments from Twitter and Facebook, we developed a custom hate speech classifier trained on an annotated corpus of 3,277 comments in Norwegian language. We identify that despite representing a small share of the total comments, hate speech content has increased over time. In an effort to understand the social network characteristics of hate speech content, we delve deeper into Twitter conversations as we can more easily identify how this content is spread. We develop network metrics to assess the prevalence, distribution, and diffusion of hateful content. The findings reveal that regardless of the number of users or tweets in a conversation, the volume of hateful content tends to remain constant. Furthermore, a small fraction of users contribute disproportionately to the dissemination of hate speech, with most conversations being limited in participant diversity. These results contribute to the growing field of computational social science by offering a novel methodology for studying hate speech in under-resourced languages and suggesting that mitigating hate speech may be possible through targeted network interventions rather than content removal alone.

• Djenouri, Youcef; Belmecheri, Nassim; Michalak, Tomasz; Dubiński, Jan; Belbachir, Nabil & Yazidi, Anis (2025). Learning Graph Representation of Agent Diffusers. In y Das, Sanmay; Nowé, Ann & Vorobeychik, Yevgeniy (Ed.), AAMAS '25: Proceedings of the 24th International Conference on Autonomous Agents and Multiagent Systems. Association for Computing Machinery (ACM). ISSN 9798400714269. p. 620–629. doi: https:/dl.acm.org/doi/10.5555/3709347.3743578. Full text in Research Archive Show summary

  Diffusion-based generative models have significantly advanced text-to-image synthesis, demonstrating impressive text comprehension and zero-shot generalization. These models refine images from random noise based on textual prompts, with initial reliance on text input shifting towards enhanced visual fidelity over time. This transition suggests that static model parameters might not optimally address the distinct phases of generation. We introduce LGR-AD (Learning Graph Representation of Agent Diffusers), a novel multi-agent system designed to improve adaptability in dynamic computer vision tasks. LGR-AD models the generation process as a distributed system of interacting agents, each representing an expert sub-model. These agents dynamically adapt to varying conditions and collaborate through a graph neural network that encodes their relationships and performance metrics. Our approach employs a coordination mechanism based on top-k maximum spanning trees, optimizing the generation process. Each agent's decision-making is guided by a meta-model that minimizes a novel loss function, balancing accuracy and diversity. Theoretical analysis and extensive empirical evaluations show that LGR-AD outperforms traditional diffusion models across various benchmarks, highlighting its potential for scalable and flexible solutions in complex image generation tasks. Code can be found at: https://github.com/YousIA/LGR_AD.

• Krejcar, Ondrej; Yazidi, Anis; Agarwal, Gaurav & Namazi, Hamidreza (2025). COMPLEXITY-BASED ANALYSIS OF THE VARIATIONS IN BRAIN ACTIVITY IN SHORT DEEP BREATHING. Fractals. ISSN 0218-348X. 33(9). doi: 10.1142/S0218348X25500707. Full text in Research Archive
• Silva, Pedro Rego Lencastre e; Bystrik, Yurii; Yazidi, Anis; Denysov, Sergiy & Lind, Pedro (2025). Dynamical law behind eye movements: Distinguishing between Levy and intermittent strategies. Physical Review Research (PRResearch). 7. doi: 10.1103/2bzm-t9k1. Full text in Research Archive
• Presacan, Oriana; Ojha, Jaya; Yazidi, Anis; Monteiro, Eric & Lind, Pedro G. (2025). A Comprehensive Review of Explainable AI in Deep Learning Algorithms for EEG Analysis. ACM Transactions on Computing for Healthcare (HEALTH). ISSN 2691-1957. doi: 10.1145/3785409. Full text in Research Archive Show summary

  While deep learning techniques are nowadays a powerful field to automatically learn and perform accurate EEG data classification in the clinical context, they still lack wide acceptance within the medical and health research community. This lack of trust is associated with the high complexity deep learning algorithms typically have which contributes to a low level of interpretability of the outcome and predictions. This survey aims to provide a comprehensive discussion of the latest advancements in deep learning models applied to EEG analysis, while also emphasizing research in explainable AI within this domain. It explores commonly used algorithms in EEG analysis, their main application areas, and the insights provided by XAI. Moreover, the survey addresses current limitations, such as the evaluation of XAI methods and the need for clinical validation, as well as ongoing challenges in this field. One critical insight from this review is the relative paucity of clinical evaluations of the rich stock of proposed techniques and methods for AI explainability. By showcasing various applications and breakthroughs in EEG analysis facilitated by XAI, the survey underscores the potential of these technologies to revolutionize neurological diagnosis and treatment, paving the way for wider acceptance and implementation in clinical settings.

• Gorosito, Martin; Berg, Bjørnar; Yazidi, Anis; Hermansen, Åsmund; Øiestad, Britt Elin & Grotle, Margreth [Show all 8 contributors for this article] (2025). Disability pension during emerging adulthood: Insights from the young-HUNT study on psychological distress, chronic pain, and policy reform. International Journal of Medical Informatics. ISSN 1386-5056. 207. doi: 10.1016/j.ijmedinf.2025.106216. Full text in Research Archive Show summary

  Introduction: Disability pension (DP) among young adults has steadily increased in Norway, with chronic pain and psychological distress among the main causes. This study examined the association between co-occurring psychological distress and chronic pain in adolescence and the risk of DP during emerging adulthood, before and after the Norwegian Labour and Welfare Administration (NAV) reform. Machine learning survival models were also developed to predict early DP and identify consistent risk factors across cohorts. Methods: Data from two general population surveys in 1995–1997 (Young-HUNT1) and 2006–2008 (Young-HUNT3) were linked to public registry data for DP, with a mean follow-up of 12 years. These represent populations pre- and post-NAV reform, respectively. Crude and adjusted Cox proportional hazards models tested associations between chronic pain, psychological distress and their co-occurrence with early DP. Machine learning survival models were developed using common predictors across cohorts and validated temporally. SurvSHAP(t) and hazard ratios were used for explainable risk factor analysis. Results: Co-occurring psychological distress and chronic pain were associated with increased early DP risk in Young-HUNT3 (HR = 2.25, [95 % CI 1.62 to 3.14]) but not in Young-HUNT1. Tree-based machine learning models showed strong predictive performance and generalizability over time, with ExtraSurvivalTrees achieving a validation c-index of 0.75 [95 % CI 0.74–0.76]. Consistent risk factors included low social integration, limited physical activity, and low parental education in both cohorts, while additional factors post-reform included low self-esteem, poor self-rated health, and negative school experiences. Conclusion: Adolescents with co-occurring psychological distress and chronic pain face an elevated risk of early DP. Early identification and support are crucial, as risk factors extend beyond medical conditions. Findings also suggest that NAV policy changes may have influenced DP receipt rates among young adults.

• Belhadi, Asma; Djenouri, Youcef; Lind, Pedro & Yazidi, Anis (2025). Hybrid Visibility Graph and Long Short Term Memory for Schizophrenia Detection. Proceedings of the International Joint Conference on Neural Networks. ISSN 2161-4393. p. 1–8. doi: 10.1109/ijcnn64981.2025.11228894. Full text in Research Archive Show summary

  Schizophrenia is a complex neuropsychiatric disorder that affects cognitive function and brain activity. Electroencephalography (EEG) has emerged as a valuable tool for detecting schizophrenia-related neural patterns, but accurate classification remains a challenge due to the intricate nature of EEG signals. In this study, we propose a Hybrid Visibility Graph and Long Short-Term Memory (VG-LSTM) framework for schizophrenia detection. Our approach transforms EEG time series into graph structures using visibility graphs (VGs) to capture the underlying topological properties of brain activity. We then employ LSTM networks to model sequential dependencies, effectively integrating both structural and sequential information for robust classification. Experimental results on publicly available schizophrenia EEG datasets demonstrate that our VG-LSTM framework achieves superior performance compared to conventional deep learning approaches. The results highlight the potential of combining graph-theoretic and deep sequential modeling techniques for EEG-based neuropsychiatric disorder detection. The full code of this research work is available on https://github.com/YousIA/VG-LSTM

• Khadka, Rabindra; Sengupta, Poushali; Lind, Pedro & Yazidi, Anis (2025). Quantization of Vision Transformer-Based Model for Real-Time EEG Classification. In Azab, Abdulrahman & Malki, Tomasz (Ed.), Nordic e-Infrastructure Tomorrow 6th Nordic e-Infrastructure Collaboration Conference NeIC 2024, Tallinn, Estonia, May 27–29, 2024 Proceedings. Springer Nature. ISSN 9783031862403. p. 17–27. doi: 10.1007/978-3-031-86240-3_2. Full text in Research Archive Show summary

  Abstract Electroencephalography (EEG) is a non-invasive and cost-effective tool for capturing brain signals. However, existing approaches based on deep learning models for classifying EEG signals are primarily trained on large EEG datasets and require extensive computational resources. This poses significant challenges for real-time processing, particularly in resource-constrained environments such as wearable devices and edge computing. Our work addresses these challenges by exploring quantization techniques to optimize a vision transformer model for real-time EEG classification. We apply post-training quantization to the proposed Vision Transformer (ViT) model, termed the Brain Signal Vision Transformer (BSVT), to reduce computational overhead while maintaining high performance. The BSVT model, trained on the ex3 HPC system, efficiently handles large datasets. We comprehensively analyze the trade-offs between model accuracy and computational efficiency. Specifically, we demonstrate the use of dynamic quantization on a ViT-based model for EEG classification, achieving a 2.49x reduction in model size and a 1.46x improvement in inference speed. These findings underscore the potential of quantization techniques and distributed computing to revolutionize the adoption of AI in healthcare.

• Ojha, Jaya; Presacan, Oriana; Lind, Pedro G.; Monteiro, Eric & Yazidi, Anis (2025). Navigating Uncertainty: A User-Perspective Survey of Trustworthiness of AI in Healthcare. ACM Transactions on Computing for Healthcare (HEALTH). ISSN 2691-1957. 6(3). doi: https:/dl.acm.org/doi/10.1145/3716317. Full text in Research Archive Show summary

  This article offers an extensive survey of one of the fundamental aspects of the trustworthiness of AI in healthcare, namely uncertainty, focusing on the large panoply of recent studies addressing the connection between uncertainty, AI, and healthcare. The concept of uncertainty is a recurring theme across multiple disciplines, with varying focuses and approaches. Here, we focus on the diverse nature of uncertainty in medical applications, emphasizing the importance of quantifying uncertainty in model predictions and its advantages in specific clinical settings. Questions that emerge in this context range from the guidelines for AI integration in the healthcare domain to the ethical deliberations and their compatibility with cutting-edge AI research. Together with a description of the main specific works in this context, we also discuss that, as medicine evolves and introduces novel sources of uncertainty, there is a need for more versatile uncertainty quantification methods to be developed collaboratively by researchers and healthcare professionals. Finally, we acknowledge the limitations of current uncertainty quantification methods in addressing the different facets of uncertainty within the medical domain. In particular, we identify from this survey a relative paucity of approaches that focus on the user’s perception of uncertainty and accordingly of trustworthiness.

• Belhadi, Asma; Yazidi, Anis; Lind, Pedro & Djenouri, Youcef (2025). EEG Data Classification: Review and Taxonomy. ACM Transactions on Computing for Healthcare (HEALTH). ISSN 2691-1957. 6(4), p. 1–29. doi: 10.1145/3742795. Full text in Research Archive Show summary

  EEG data classification plays a pivotal role in understanding brain activity and its applications in various domains. Deep learning has emerged as a powerful paradigm for automatically learning complex patterns from raw data, eliminating the need for manual feature extraction. However, in the context of medical data, and in particular for EEG analysis, the use of deep learning approaching while having been very successful is not being included in medical diagnosis routines, yet. The aim of this survey is twofold. On one side, it provides a comprehensive overview of the current state-of-the-art in EEG data classification, with a specific focus on the use of deep learning techniques. On the other side, it also addresses the clinician community, explaining the power and trustfulness of such new approaches. The survey begins with an introduction highlighting the limitations of traditional model-based approaches and the potential of deep learning in EEG data classification. The fundamental principles and architectures of deep learning models are presented, including convolutional neural networks (CNNs), recurrent neural networks (RNNs), and graph convolution neural networks (GCNNs) that have been successfully applied to EEG data classification tasks. A detailed review and analysis of existing literature on deep learning-based EEG data classification are provided, categorizing the studies based on the type of the input data, e.g., sequences, images, graphs, or multi-modalities. We also discuss about the existing tools and technologies for EEG data classification and highlights the challenges and limitations associated with deep learning in EEG data classification, including limited data availability, interpretability of deep models, and bias mitigation. Potential solutions and ongoing research efforts to overcome these challenges are explored, providing insights into the future directions of this field. This survey serves as a valuable resource for researchers, practitioners, and healthcare professionals involved in EEG data classification. It provides an extensive understanding of the advancements, challenges, and potential applications of deep learning techniques in this domain, guiding further research and development of accurate and interpretable approaches for EEG data analysis and interpretation.

• Acharya, Sushil; Fabian, Karl; Yazidi, Anis & Westeng, Kjetil (2025). Deep Hierarchical Graph Correlation: A Two-Stage Approach to Well-Log Alignment Using CNNs and Dynamic Programming. Journal of Marine Science and Engineering (JMSE). 14(1), p. 66–66. doi: 10.3390/jmse14010066. Full text in Research Archive Show summary

  Precise depth alignment of well logs is essential for reliable subsurface characterization, enabling accurate correlation of geological features across multiple wells. This study presents the Deep Hierarchical Graph Correlator (DHGC), a two-stage deep learning framework for scalable and automated well-log depth alignment. DHGC aligns a target log to a reference log by comparing fixed-size windows extracted from both signals. In the first stage, a one-dimensional convolutional neural network (1D CNN) trained on 177,026 triplets using triplet-margin loss learns discriminative embeddings of gamma-ray (GR) log windows from eight Norwegian North Sea wells. In the second stage, a feedforward scoring network evaluates embedded window pairs to estimate local similarity. Dynamic programming then computes the optimal nonlinear warping path from the resulting cost matrix. The feature extractor achieved 99.6% triplet accuracy, and the scoring network achieved 98.93% classification accuracy with an ROC-AUC of 0.9971. Evaluation on 89 unseen GR log pairs demonstrated that DHGC improves the mean Pearson correlation coefficient from 0.35 to 0.91, with successful alignment in 88 cases (98.9%). DHGC achieved an 8.2× speedup over DTW (3.16 s versus 25.83 s per log pair). While DTW achieves a higher mean correlation (0.96 versus 0.91), DHGC avoids singularity artifacts and exhibits lower variability in distance metrics than CC, suggesting improved robustness and scalability for well-log synchronization.

• Szczepanski, Tomasz Szymon; Omland, Petter Moe; Dunker, Øystein; Sand, Trond Halfdan; Tannemaat, Martijn & Reijntjes, Robert [Show all 11 contributors for this article] (2025). Deriving reference limits from historical data – A comparison of four novel methods. Clinical Neurophysiology. ISSN 1388-2457. 182. doi: 10.1016/j.clinph.2025.2111451. Full text in Research Archive Show summary

  Objective To compare four novel methods – extrapolated norms (E-norms), extrapolated reference values (E-Ref), multivariable extrapolated reference values (MeRef) and mixture model clustering (MMC) – for obtaining reference limits for nerve conduction studies (NCS) from historical data containing both normal and abnormal studies. Methods Reference limits for 29 commonly clinically used NCS measurements were calculated from a historical database containing measurements from 24 618 patients using E-norms, E-ref, MeRef and MMC. The resulting reference limits were compared to reference limits for NCS calculated from 680 healthy subjects using Youden’s J statistics. Results Except for distal latencies, E-norms produced reference limits with the highest Youden’s J statistics with higher sensitivity, but lower specificity. E-Ref, MeRef and MMC produced reference limits with high specificity, but lower sensitivity than E-norms. Conclusions There are substantial differences in the performance of E-norms, E-Ref, MeRef and MMC. A dynamic approach, where the methods used are changed based on the type of NCS measurement and the amount of historical data available, may yield the highest accuracy. Significance When combining the different novel methods, it is possible to create clinically useful reference limits using historical data.

• Gorosito, Martin; Haugerud, Hårek; Berg, Bjørnar; Yazidi, Anis; Øiestad, Britt Elin & Sivertsen, Børge [Show all 8 contributors for this article] (2025). Chronic pain, psychological distress, and their co-occurrence in Norwegian young adults: Insights from machine learning and explainable AI. International Journal of Medical Informatics. ISSN 1386-5056. 205. doi: 10.1016/j.ijmedinf.2025.106108. Full text in Research Archive Show summary

  Introduction Chronic pain and psychological distress significantly impact young adults' health, yet little is known about their associated factors when they co-occur. This study applied machine learning classification models and explainable AI to uncover associations across diverse factors with chronic pain, psychological distress, and their co-occurrence in young adults. It also examined how associations vary when these conditions occur individually versus when they co-occur. Variable clusters were also obtained to identify latent subgroups. Methods Cross-sectional data from the Norwegian Students' Health and Well-being Study (SHoT2018) were utilised. Three classification models (Logistic Regression, Tsetlin Machine, CatBoost) were developed separately for three mutually exclusive groups (chronic pain only, psychological distress only, and their co-occurrence), against a healthy group. Model performance was assessed using Area Under the Curve (AUC) and internal-external cross-validation using geographical regions as folds. Shapley values and odds ratios were applied to identify factors associated with conditions, with associations compared across them. Clause rules were obtained to search for variable clusters. Results A total of 34,469 participants (mean age 22.9 years; 67.4% female) were included. Chronic pain, psychological distress, and co-occurrence were reported by 27.2%, 12.7%, and 26.7%, respectively. Models performed consistently across regions (AUC 0.70–0.72 for chronic pain, 0.92–0.93 for psychological distress, and 0.94–0.96 for co-occurrence. Explainable AI analyses identified females, perceived health status, and sleep disorders as factors associated with all conditions. Associations shifted when conditions co-occurred versus individually, with some negating each other while others combined, revealing distinct patterns for the co-occurring group. Conclusion The findings underscore the importance of studying these conditions both individually and simultaneously. Machine learning and explainable AI revealed unique associations that can help with early identification of subjects at risk.

• Belaid, Mohamed-Bachir; Sharma, Jivitesh; Lei, Jiao; Granmo, Ole-Christoffer; Andersen, Per-Arne & Yazidi, Anis (2025). Generalized Convergence Analysis of Tsetlin Automaton Based Algorithms: A Probabilistic Approach to Concept Learning. In Walsh, Toby; Shah, Julie & Kolter, Zico (Ed.), Proceedings of the Thirty-Ninth AAAI Conference on Artificial Intelligence. AAAI Press. ISSN 9781577358978. p. 15524–15531. doi: 10.1609/aaai.v39i15.33704. Full text in Research Archive
• Khadka, Rabindra; Lind, Pedro; Yazidi, Anis & Belhadi, Asma (2025). DREAMS: A python framework for training deep learning models on EEG data with model card reporting for medical applications. SoftwareX. 30. doi: 10.1016/j.softx.2025.102140. Full text in Research Archive
• Bhandari, Shailendra; Silva, Pedro Rego Lencastre e; Mathema, Rujeena; Szorkovszky, Alexander; Yazidi, Anis & Lind, Pedro (2025). Modeling eye gaze velocity trajectories using GANs with spectral loss for enhanced fidelity. Scientific Reports. 15(1). doi: 10.1038/s41598-025-05286-5. Full text in Research Archive Show summary

  Accurate modeling of eye gaze dynamics is essential for advancement in human-computer interaction, neurological diagnostics, and cognitive research. Traditional generative models like Markov models often fail to capture the complex temporal dependencies and distributional nuance inherent in eye gaze trajectories data. This study introduces a Generative Adversarial Network (GAN) framework employing Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) generators and discriminators to generate high-fidelity synthetic eye gaze velocity trajectories. We conducted a comprehensive evaluation of four GAN architectures: CNN-CNN, LSTM-CNN, CNN-LSTM, and LSTM-LSTM–trained under two conditions: using only adversarial loss ($$L_G$$) and using a weighted combination of adversarial and spectral losses. Our findings reveal that the LSTM-CNN architecture trained with this new loss function exhibits the closest alignment to the real data distribution, effectively capturing both the distribution tails and the intricate temporal dependencies. The inclusion of spectral regularization significantly enhances the GANs’ ability to replicate the spectral characteristics of eye gaze movements, leading to a more stable learning process and improved data fidelity. Comparative analysis with a Hidden Markov Model (HMM) optimized to four hidden states further highlights the advantages of the LSTM-CNN GAN. Statistical metrics show that the HMM-generated data significantly diverges from the real data in terms of mean, standard deviation, skewness, and kurtosis. In contrast, the LSTM-CNN model closely matches the real data across these statistics, affirming its capacity to model the complexity of eye gaze dynamics effectively. These results position the spectrally regularized LSTM-CNN GAN as a robust tool for generating synthetic eye gaze velocity data with high fidelity. Its ability to accurately replicate both the distributional and temporal properties of real data holds significant potential for applications in simulation environments, training systems, and the development of advanced eye-tracking technologies, ultimately contributing to more naturalistic and responsive human-computer interactions.

• Belhadi, Asma; Lind, Pedro; Djenouri, Youcef & Yazidi, Anis (2025). Enhanced visibility graph for EEG classification. Frontiers in Neuroscience. ISSN 1662-4548. 19. doi: 10.3389/fnins.2025.1541062. Full text in Research Archive Show summary

  Electroencephalography (EEG) holds immense potential for decoding complex brain patterns associated with cognitive states and neurological conditions. In this paper, we propose an end-to-end framework for EEG classification that integrates power spectral density (PSD) and visibility graph (VG) features together with deep learning (DL) techniques. Our framework offers a holistic approach for capturing both frequency-domain characteristics and temporal dynamics of EEG signals. We evaluate four DL architectures, namely multilayer perceptron (MLP), long short-term memory (LSTM) networks, InceptionTime and ChronoNet, applied to several datasets and in different experimental conditions. Results demonstrate the efficacy of our framework in accurately classifying EEG data, in particular when using VG features. We also shed new light on the relative strengths and limitations of different feature extraction methods and DL architectures in the context of EEG classification. Our work contributes to advancing EEG analysis and facilitating the development of more accurate and reliable EEG-based systems for neuroscience and beyond. The full code of this research work is available on https://github.com/asmab89/VisibilityGraphs.git.

• Elmisadr, Negar; Belaid, Mohamed-Bachir & Yazidi, Anis (2025). Stochastic and deterministic processes in Asymmetric Tsetlin Machine. Frontiers in Artificial Intelligence. 8. doi: 10.3389/frai.2025.1377944. Full text in Research Archive Show summary

  This paper introduces a new approach to enhance the decision-making capabilities of the Tsetlin Machine (TM) through the Stochastic Point Location (SPL) algorithm and the Asymmetric Steps technique. We incorporate stochasticity and asymmetry into the TM's process, along with a decaying normal distribution function that improves adaptability as it converges toward zero over time. We present two methods: the Asymmetric Probabilistic Tsetlin (APT) Machine, influenced by random events, and the Asymmetric Tsetlin (AT) Machine, which transitions from probabilistic to deterministic states. We evaluate these methods against traditional machine learning algorithms and classical Tsetlin (CT) machines across various benchmark datasets. Both AT and APT demonstrate competitive performance, with the AT model notably excelling, especially in complex datasets.

• Aryan, Ramtin; Yazidi, Anis; Bouhoula, Adel & Engelstad, Paal E. (2025). A formal technique for automatic resolution of OpenFlow anomalies. International Journal of Information Security. ISSN 1615-5262. 24. doi: 10.1007/s10207-025-01035-x. Full text in Research Archive Show summary

  Conflicts between network policies are frequent in today’s computer networks due to the increasing complexity of network configuration. Resolving policy anomalies usually requires network administrator intervention, which is a time-intensive and error-prone process. This paper presents inference systems for the automatic resolution of OpenFlow anomalies. The approach uses high-level policies to detect conflict correction without policy violations. Our approach is fully automated and does not require interaction with the network administrator. Although there is a multitude of research papers on detecting anomalies in SDN, research to correct those anomalies in an automatic manner is very scarce, if not non-existent. We have formally proven the soundness and completeness of our inference systems. Furthermore, we provide experimental results based on real-life network configurations involving more than 1200 rules. The detection and correction processes exhibit very low computation overhead, in the order of milliseconds, when parallelization is used.

• Jha, Debesh; Durak, Gorkem; Sharma, Vanshali; Keles, Elif; Cicek, Vedat & Zhang, Zheyuan [Show all 15 contributors for this article] (2025). A Conceptual Framework for Applying Ethical Principles of AI to Medical Practice. Bioengineering. 12(2). doi: 10.3390/bioengineering12020180. Full text in Research Archive Show summary

  Artificial Intelligence (AI) is reshaping healthcare through advancements in clinical decision support and diagnostic capabilities. While human expertise remains foundational to medical practice, AI-powered tools are increasingly matching or exceeding specialist-level performance across multiple domains, paving the way for a new era of democratized healthcare access. These systems promise to reduce disparities in care delivery across demographic, racial, and socioeconomic boundaries by providing high-quality diagnostic support at scale. As a result, advanced healthcare services can be affordable to all populations, irrespective of demographics, race, or socioeconomic background. The democratization of such AI tools can reduce the cost of care, optimize resource allocation, and improve the quality of care. In contrast to humans, AI can potentially uncover complex relationships in the data from a large set of inputs and generate new evidence-based knowledge in medicine. However, integrating AI into healthcare raises several ethical and philosophical concerns, such as bias, transparency, autonomy, responsibility, and accountability. In this study, we examine recent advances in AI-enabled medical image analysis, current regulatory frameworks, and emerging best practices for clinical integration. We analyze both technical and ethical challenges inherent in deploying AI systems across healthcare institutions, with particular attention to data privacy, algorithmic fairness, and system transparency. Furthermore, we propose practical solutions to address key challenges, including data scarcity, racial bias in training datasets, limited model interpretability, and systematic algorithmic biases. Finally, we outline a conceptual algorithm for responsible AI implementations and identify promising future research and development directions.

• Sahu, Geet; Seal, Ayan; Yazidi, Anis & Krejcar, Ondrej (2024). A Dual-Channel Dehaze-Net for Single Image Dehazing in Visual Internet of Things Using PYNQ-Z2 Board. IEEE Transactions on Automation Science and Engineering. ISSN 1545-5955. 21(1), p. 305–319. doi: 10.1109/TASE.2022.3217801. Full text in Research Archive
• Upreti, Ramesh; Lind, Pedro; Elmokashfi, Ahmed Mustafa Abdalla & Yazidi, Anis (2024). Trustworthy machine learning in the context of security and privacy. International Journal of Information Security. ISSN 1615-5262. 23. doi: 10.1007/s10207-024-00813-3. Full text in Research Archive Show summary

  Artificial intelligence-based algorithms are widely adopted in critical applications such as healthcare and autonomous vehicles. Mitigating the security and privacy issues of AI models, and enhancing their trustworthiness have become of paramount importance. We present a detailed investigation of existing security, privacy, and defense techniques and strategies to make machine learning more secure and trustworthy. We focus on the new paradigm of machine learning called federated learning, where one aims to develop machine learning models involving different partners (data sources) that do not need to share data and information with each other. In particular, we discuss how federated learning bridges security and privacy, how it guarantees privacy requirements of AI applications, and then highlight challenges that need to be addressed in the future. Finally, after having surveyed the high-level concepts of trustworthy AI and its different components and identifying present research trends addressing security, privacy, and trustworthiness separately, we discuss possible interconnections and dependencies between these three fields. All in all, we provide some insight to explain how AI researchers should focus on building a unified solution combining security, privacy, and trustworthy AI in the future.

• Gambin, Angel; Yazidi, Anis; Vasilakos, Athanasios; Haugerud, Hårek & Djenouri, Youcef (2024). Deepfakes: current and future trends. Artificial Intelligence Review. ISSN 0269-2821. 57(3), p. 1–32. doi: 10.1007/s10462-023-10679-x. Full text in Research Archive Show summary

  Advances in Deep Learning (DL), Big Data and image processing have facilitated online disinformation spreading through Deepfakes. This entails severe threats including public opinion manipulation, geopolitical tensions, chaos in financial markets, scams, defamation and identity theft among others. Therefore, it is imperative to develop techniques to prevent, detect, and stop the spreading of deepfake content. Along these lines, the goal of this paper is to present a big picture perspective of the deepfake paradigm, by reviewing current and future trends. First, a compact summary of DL techniques used for deepfakes is presented. Then, a review of the fight between generation and detection techniques is elaborated. Moreover, we delve into the potential that new technologies, such as distributed ledgers and blockchain, can offer with regard to cybersecurity and the fight against digital deception. Two scenarios of application, including online social networks engineering attacks and Internet of Things, are reviewed where main insights and open challenges are tackled. Finally, future trends and research lines are discussed, pointing out potential key agents and technologies.

• Khan, Haroon; Khadka, Rabindra; Sultan, Malik Shahid; Yazidi, Anis; Ombao, Hernando & Mirtaheri, Peyman (2024). Unleashing the potential of fNIRS with machine learning: classification of fine anatomical movements to empower future brain-computer interface. Frontiers in Human Neuroscience. 18. doi: 10.3389/fnhum.2024.1354143. Full text in Research Archive Show summary

  In this study, we explore the potential of using functional near-infrared spectroscopy (fNIRS) signals in conjunction with modern machine-learning techniques to classify specific anatomical movements to increase the number of control commands for a possible fNIRS-based brain-computer interface (BCI) applications. The study focuses on novel individual finger-tapping, a well-known task in fNIRS and fMRI studies, but limited to left/right or few fingers. Twenty-four right-handed participants performed the individual finger-tapping task. Data were recorded by using sixteen sources and detectors placed over the motor cortex according to the 10-10 international system. The event's average oxygenated Δ HbO and deoxygenated Δ HbR hemoglobin data were utilized as features to assess the performance of diverse machine learning (ML) models in a challenging multi-class classification setting. These methods include LDA, QDA, MNLR, XGBoost, and RF. A new DL-based model named “Hemo-Net” has been proposed which consists of multiple parallel convolution layers with different filters to extract the features. This paper aims to explore the efficacy of using fNRIS along with ML/DL methods in a multi-class classification task. Complex models like RF, XGBoost, and Hemo-Net produce relatively higher test set accuracy when compared to LDA, MNLR, and QDA. Hemo-Net has depicted a superior performance achieving the highest test set accuracy of 76%, however, in this work, we do not aim at improving the accuracies of models rather we are interested in exploring if fNIRS has the neural signatures to help modern ML/DL methods in multi-class classification which can lead to applications like brain-computer interfaces. Multi-class classification of fine anatomical movements, such as individual finger movements, is difficult to classify with fNIRS data. Traditional ML models like MNLR and LDA show inferior performance compared to the ensemble-based methods of RF and XGBoost. DL-based method Hemo-Net outperforms all methods evaluated in this study and demonstrates a promising future for fNIRS-based BCI applications.

• Jouini, Oumayma; Aoueileyine, Mohamed Ould-Elhassen; Sethom, Kaouthar & Yazidi, Anis (2024). Wheat Leaf Disease Detection: A Lightweight Approach with Shallow CNN Based Feature Refinement. AgriEngineering. 6(3), p. 2001–2022. doi: 10.3390/agriengineering6030117. Full text in Research Archive Show summary

  Improving agricultural productivity is essential due to rapid population growth, making early detection of crop diseases crucial. Although deep learning shows promise in smart agriculture, practical applications for identifying wheat diseases in complex backgrounds are limited. In this paper, we propose CropNet, a hybrid method that utilizes Red, Green, and Blue (RGB) imaging and a transfer learning approach combined with shallow convolutional neural networks (CNN) for further feature refinement. To develop our customized model, we conducted an extensive search for the optimal deep learning architecture. Our approach involves freezing the pre-trained model for feature extraction and adding a custom trainable CNN layer. Unlike traditional transfer learning, which typically uses trainable dense layers, our method integrates a trainable CNN, deepening the architecture. We argue that pre-trained features in transfer learning are better suited for a custom shallow CNN followed by a fully connected layer, rather than being fed directly into fully connected layers. We tested various architectures for pre-trained models including EfficientNetB0 and B2, DenseNet, ResNet50, MobileNetV2, MobileNetV3-Small, and Inceptionv3. Our approach combines the strengths of pre-trained models with the flexibility of custom architecture design, offering efficiency, effective feature extraction, customization options, reduced overfitting, and differential learning rates. It distinguishes itself from classical transfer learning techniques, which typically fine-tune the entire pre-trained network. Our aim is to provide a lightweight model suitable for resource-constrained environments, capable of delivering outstanding results. CropNet achieved 99.80% accuracy in wheat disease detection with reduced training time and computational cost. This efficient performance makes CropNet promising for practical implementation in resource-constrained agricultural settings, benefiting farmers and enhancing production.

• Ojha, Jaya; Haugerud, Hårek; Yazidi, Anis & Lind, Pedro (2024). Exploring Interpretable AI Methods for ECG Data Classification. In for, Computing Mach Association (Eds.), ICDAR 2024 Proceedings. Association for Computing Machinery (ACM). ISSN 9798400705496. p. 11–18. doi: 10.1145/3643488.3660294. Full text in Research Archive Show summary

  We address ECG data classification, using methods from explainable artificial intelligence (XAI). In particular, we focus on the extended performance of the ST-CNN-5 model compared to established models. The model showcases slight improvement in accuracy suggesting the potential of this new model to provide more reliable predictions compared to other models. However, lower values of the specificity and area-under-curve metrics highlight the need to thoroughly evaluate the strengths and weaknesses of the extended model compared to other models. For the interpretability analysis, we use Shapley Additive Explanations (SHAP), Gradient-weighted Class Activation Mapping (GradCAM), and Local Interpretable Model-agnostic Explanations (LIME) methods. In particular, we show that the new model exhibits improved explainability in its GradCAM explanations compared to the former model. SHAP effectively highlights crucial ECG features, better than GradCAM and LIME. The latter methods exhibit inferior performance, particularly in capturing nuanced patterns associated with certain cardiac conditions. By using distinctive methods in the interpretability analysis, we provide a systematic discussion about which ECG features are better - or worse - uncovered by each method.

• Silva, Pedro Rego Lencastre e; Yazidi, Anis & Lind, Pedro (2024). Modeling Wind-Speed Statistics beyond the Weibull Distribution. Energies. 17(11). doi: 10.3390/en17112621. Full text in Research Archive Show summary

  While it is well known that the Weibull distribution is a good model for wind-speed measurements and can be explained through simple statistical arguments, how such a model holds for shorter time periods is still an open question. In this paper, we present a systematic investigation of the accuracy of the Weibull distribution to wind-speed measurements, in comparison with other possible “cousin” distributions. In particular, we show that the Gaussian distribution enables one to predict wind-speed histograms with higher accuracy than the Weibull distribution. Two other good candidates are the Nakagami and the Rice distributions, which can be interpreted as particular cases of the Weibull distribution for particular choices of the shape and scale parameters. These findings hold not only when predicting next-point values of the wind speed but also when predicting the wind energy values. Finally, we discuss such findings in the context of wind power forecasting and monitoring for power-grid assessment.

• Tveter, Mats; Tveitstøl, Thomas; Hatlestad-Hall, Christoffer; Teseyra, Ana Silvina Pérez; Taubøll, Erik & Yazidi, Anis [Show all 8 contributors for this article] (2024). Advancing EEG prediction with deep learning and uncertainty estimation. Brain Informatics. ISSN 2198-4026. 11(1). doi: 10.1186/s40708-024-00239-6. Full text in Research Archive Show summary

  Deep Learning (DL) has the potential to enhance patient outcomes in healthcare by implementing proficient systems for disease detection and diagnosis. However, the complexity and lack of interpretability impede their widespread adoption in critical high‑stakes predictions in healthcare. Incorporating uncertainty estimations in DL systems can increase trustworthiness, providing valuable insights into the model’s confidence and improving the explanation of predictions. Additionally, introducing explainability measures, recognized and embraced by healthcare experts, can help address this challenge. In this study, we investigate DL models’ ability to predict sex directly from electro‑encephalography (EEG) data. While sex prediction have limited direct clinical application, its binary nature makes it a valuable benchmark for optimizing deep learning techniques in EEG data analysis. Furthermore, we explore the use of DL ensembles to improve performance over single models and as an approach to increase interpretability and performance through uncertainty estimation. Lastly, we use a data‑driven approach to evaluate the relationship between frequency bands and sex prediction, offering insights into their relative importance. InceptionNetwork, a single DL model, achieved 90.7% accuracy and an AUC of 0.947, and the best‑performing ensemble, combining variations of InceptionNetwork and EEGNet, achieved 91.1% accuracy in predicting sex from EEG data using five‑fold cross‑validation. Uncertainty estimation through deep ensembles led to increased prediction performance, and the models were able to classify sex in all frequency bands, indicating sex‑specific features across all bands

• Karmous, Nader; Dhiab, Yassmine Ben; Aoueileyine, Mohamed Ould-Elhassen; Youssef, Neji; Bouallegue, Ridha & Yazidi, Anis (2024). Deep learning approaches for protecting IoT devices in smart homes from MitM attacks. Frontiers in Computer Science. 6. doi: 10.3389/fcomp.2024.1477501. Full text in Research Archive
• Brathovde, Kristian; Djenouri, Youcef; Yazidi, Anis & Srivastava, Gautam (2024). Grouping Interesting Patterns for Understanding Customer Behaviors. IEEE Transactions on Computational Social Systems. doi: 10.1109/TCSS.2024.3492094. Full text in Research Archive
• Yazidi, Anis; Hammer, Hugo Lewi & Leslie, David S. (2024). A Two-Timescale Learning Automata Solution to the Nonlinear Stochastic Proportional Polling Problem. IEEE Transactions on Systems, Man & Cybernetics. Systems. ISSN 2168-2216. 54(12), p. 7158–7169. doi: 10.1109/TSMC.2024.3414832. Full text in Research Archive
• Khan, Mehak; Mello, Gustavo Borges Moreno e; Habib, Laurence; Engelstad, Paal & Yazidi, Anis (2024). HITS-based Propagation Paradigm for Graph Neural Networks. ACM Transactions on Knowledge Discovery from Data. ISSN 1556-4681. 18(4), p. 1–23. doi: 10.1145/3638779. Full text in Research Archive Show summary

  In this article, we present a new propagation paradigm based on the principle of Hyperlink-Induced Topic Search (HITS) algorithm. The HITS algorithm utilizes the concept of a “self-reinforcing” relationship of authority-hub. Using HITS, the centrality of nodes is determined via repeated updates of authority-hub scores that converge to a stationary distribution. Unlike PageRank-based propagation methods, which rely solely on the idea of authorities (in-links), HITS considers the relevance of both authorities (in-links) and hubs (out-links), thereby allowing for a more informative graph learning process. To segregate node prediction and propagation, we use a Multilayer Perceptron in combination with a HITS-based propagation approach and propose two models: HITS-GNN and HITS-GNN+. We provided additional validation of our models’ efficacy by performing an ablation study to assess the performance of authority-hub in independent models. Moreover, the effect of the main hyper-parameters and normalization is also analyzed to uncover how these techniques influence the performance of our models. Extensive experimental results indicate that the proposed approach significantly improves baseline methods on the graph (citation network) benchmark datasets by a decent margin for semi-supervised node classification, which can aid in predicting the categories (labels) of scientific articles not exclusively based on their content but also based on the type of articles they cite.

• Abolghasemi, Roza; Viedma, Enrique Herrera; Engelstad, Paal E.; Djenouri, Youcef & Yazidi, Anis (2024). A graph neural approach for group recommendation system based on pairwise preferences. Information Fusion. ISSN 1566-2535. 107. doi: 10.1016/j.inffus.2024.102343. Full text in Research Archive Show summary

  Pairwise preference information, which involves users expressing their preferences by comparing items, plays a crucial role in decision-making and has recently found application in recommendation systems. In this study, we introduce GcPp, a clustering algorithm that leverages pairwise preference data to generate recommendations for user groups. Initially, we construct individual graphs for each user based on their pairwise preferences and utilize a graph convolutional network to predict similarities between all pairs of graphs. These predicted similarity scores form the foundation of our research. We then construct a new graph where users are nodes and the edges are weighted according to the predicted similarities. Finally, we perform clustering on the graph’s nodes (users). By evaluating various metrics, we found that employing a similarity metric based on a convolutional neural network (SimGNN) with our proposed ground truth called Top-K yielded the highest accuracy. The proposed approach is specifically designed for group recommendation systems and holds significant potential for group decision-making problems. Code is available at https://github.com/RozaAbolghasemi/Group_Recommendation_Syatem_GcPp_clustering.

• Aaby, Pernille; Biermann, Daniel; Yazidi, Anis; Mello, Gustavo Borges Moreno e & Palumbo, Fabrizio (2023). Exploring Multilingual Word Embedding Alignments in BERT Models: A Case Study of English and Norwegian. In Bramer, Max & Stahl, Frederic (Ed.), Artificial Intelligence XL: 43rd SGAI International Conference on Artificial Intelligence, AI 2023. Springer. ISSN 9783031479939. p. 47–58. doi: 10.1007/978-3-031-47994-6_4. Full text in Research Archive Show summary

  Contextual language models, such as transformers, can solve a wide range of language tasks ranging from text classification to question answering and machine translation. Like many deep learning models, the performance heavily depends on the quality and amount of data available for training. This poses a problem for low-resource languages, such as Norwegian, that can not provide the necessary amount of training data. In this article, we investigate the use of multilingual models as a step toward overcoming the data sparsity problem for minority languages. In detail, we study how words are represented by multilingual BERT models across two languages of our interest: English and Norwegian. Our analysis shows that multilingual models similarly encode English-Norwegian word pairs. The multilingual model automatically aligns semantics across languages without supervision. Additionally, our analysis also shows that embedding a word encodes information about the language to which it belongs. We, therefore, believe that in pre-trained multilingual models' knowledge from one language can be transferred to another without direct supervision and help solve the data sparsity problem for minor languages.

• Silva, Pedro Rego Lencastre e; Gjersdal, Marit Øye; Gorjão, Leonardo Rydin; Yazidi, Anis & Lind, Pedro (2023). Modern AI versus century-old mathematical models: How far can we go with generative adversarial networks to reproduce stochastic processes? Physica D : Non-linear phenomena. ISSN 0167-2789. 453. doi: 10.1016/j.physd.2023.133831. Full text in Research Archive Show summary

  The usage of generative adversarial networks (GAN)s for synthetic time-series data generation has been gaining popularity in recent years with applications from finance to music composition and processing of textual content. However, beyond their reported success, few comparisons exist with other artificial intelligence (AI) methods or standard mathematical models. Here, we test GANs performance, comparing them with a well-known mathematical model, namely a Markov chain. We implement comparative metrics based on one- and two-point statistics to evaluate the performance of each method. We find that, similarly to other AI approaches, GANs struggle to capture rare events and cross-feature relations and are unable to create synthetic faithful data. GANs are relatively successful in replicating the auto-correlation function, but they still lag significantly behind simple Markov chains. We also provide a qualitative explanation for this limitation of AI approaches.

• Djenouri, Youcef; Belhadi, Asma; Srivastava, Gautam; Lin, Jerry Chun-Wei & Yazidi, Anis (2023). Interpretable intrusion detection for next generation of Internet of Things. Computer Communications. ISSN 0140-3664. 203, p. 192–198. doi: 10.1016/j.comcom.2023.03.005. Full text in Research Archive Show summary

  This paper presents a new framework for intrusion detection in the next-generation Internet of Things. MinMax normalization strategy is used to collect and preprocess data. The Marine Predator algorithm is then used to select relevant features to be used in the learning process. The selected features are then trained with an advanced and state-of-the-art recurrent neural network that includes an attention mechanism. Finally, Shapely values are calculated to determine how much each feature contributes to the final output. The dataset NSL-KDD was used for intensive simulations. The results show the advantages of the proposed system as well as its superiority over state-of-the-art methods. In fact, the proposed solution achieved a rate of more than 94% for both true negative and true position, while the rates of the existing solutions are below 90% for the challenging NSL-KDD datasets.

• Djenouri, Youcef; Belbachir, Nabil; Michalak, Tomasz & Yazidi, Anis (2023). Shapley Deep Learning: A Consensus for General-Purpose Vision Systems. In Cho, Minsu & Ham, Bumsub (Ed.), 2023 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW). IEEE (Institute of Electrical and Electronics Engineers). ISSN 9798350307443. p. 1216–1225. doi: 10.1109/ICCVW60793.2023.00132. Full text in Research Archive Show summary

  Shapely Deep Learning (SDL) targets a new foundation for the design of general-purpose vision systems, by establishing a consensus method that facilitates self-adaptation and flexibility to deal with new computer vision tasks. Today, machine learning did not yet reach the flexible, general-purpose intelligence that biological vision has in mimicking visual descriptions and learning in general-purpose vision algorithms. Currently, each model is built using the domain knowledge of the application in question. Data scientists must consequently be well-versed in the relevant subject. This paper presents SDL as a consensus method for general-purpose intelligence without the help of a domain expert as the trained model has been developed utilizing a general deep learning approach that investigates the contribution of each model in the training process. First, several deep learning models have been trained for each image. The shapley value is then determined to compute the contribution of each subset of models in the training. The model selection is finally performed based on the shapley value and the joint model cost. Optimization of the shapley computation is also carried out by investigating the banzhaf function. We present the evaluation of the generality of SDL using the computer vision benchmarks: MNIST for Shapley value performance, ImageNet for image classification, and COCO for object detection. The results reveal the effectiveness of SDL in terms of accuracy and competitiveness of inference runtime. Concretely, SDL achieved 10%, and 8% over MViTv2 for classification, and object detection tasks, respectively.

• Khan, Mehak; Mello, Gustavo Borges Moreno e; Engelstad, Paal E.; Habib, Laurence & Yazidi, Anis (2023). HITS-GNN: A Simplified Propagation Scheme for Graph Neural Networks. In Tsumoto, Shusaku; Ohsawa, Yukio; Chen, Lei; Poel, Dirk Van den; Hu, Xiaohua; Motomura, Yoichi; Takagi, Takuya; Wu, Lingfei; Xie, Ying; Abe, Akihiro & Raghavan, Vijay (Ed.), 2022 IEEE International Conference on Big Data. IEEE (Institute of Electrical and Electronics Engineers). ISSN 9781665480451. doi: 10.1109/BigData55660.2022.10020815. Full text in Research Archive
• Belhadi, Asma; Holland, Jon-Olav; Yazidi, Anis; Srivastava, Gautam; Lin, Jerry Chun-Wei & Djenouri, Youcef (2023). BIoMT-ISeg: Blockchain internet of medical things for intelligent segmentation. Frontiers in Physiology. 13. doi: 10.3389/fphys.2022.1097204. Full text in Research Archive Show summary

  In the quest of training complicated medical data for Internet of Medical Things (IoMT) scenarios, this study develops an end-to-end intelligent framework that incorporates ensemble learning, genetic algorithms, blockchain technology, and various U-Net based architectures. Genetic algorithms are used to optimize the hyper-parameters of the used architectures. The training process was also protected with the help of blockchain technology. Finally, an ensemble learning system based on voting mechanism was developed to combine local outputs of various segmentation models into a global output. Our method shows that strong performance in a condensed number of epochs may be achieved with a high learning rate and a small batch size. As a result, we are able to perform better than standard solutions for well-known medical databases. In fact, the proposed solution reaches 95% of intersection over the union, compared to the baseline solutions where they are below 80%. Moreover, with the proposed blockchain strategy, the detected attacks reached 76%.

• Djenouri, Youcef; Yazidi, Anis; Srivastava, Gautam & Lin, Jerry Chun-Wei (2023). Blockchain: Applications, Challenges, and Opportunities in Consumer Electronics. IEEE Consumer Electronics Magazine. ISSN 2162-2248. 13(2), p. 36–41. doi: 10.1109/MCE.2023.3247911. Full text in Research Archive Show summary

  One of the most beneficial and original solutions for cybersecurity is blockchain. Although the technology has only been on the market for a short time, it has already effectively replaced economic transaction systems in a number of businesses in its early stages and is capable of modernizing various business models in numerous consumer electronic applications. It is critical for planners and decision makers to thoroughly evaluate the suitability of Blockchain for their particular use cases and commercial applications, even as it promises a secure distributed framework for sharing, exchanging, and integrating information across users and third parties. Blockchain should only be deployed if it is suitable, offers security, and provides better opportunities to drive revenue and save costs. In this paper, you will get an overview of this technology that can be used to achieve security in consumer electronic applications. We also discuss the difficulties and potential benefits of using blockchain technologies in consumer electronics applications.

• Thunold, Håvard Horgen; Riegler, Michael; Yazidi, Anis & Hammer, Hugo Lewi (2023). A Deep Diagnostic Framework Using Explainable Artificial Intelligence and Clustering. Diagnostics (Basel). 13(22). doi: 10.3390/diagnostics13223413. Full text in Research Archive Show summary

  An important part of diagnostics is to gain insight into properties that characterize a disease. Machine learning has been used for this purpose, for instance, to identify biomarkers in genomics. However, when patient data are presented as images, identifying properties that characterize a disease becomes far more challenging. A common strategy involves extracting features from the images and analyzing their occurrence in healthy versus pathological images. A limitation of this approach is that the ability to gain new insights into the disease from the data is constrained by the information in the extracted features. Typically, these features are manually extracted by humans, which further limits the potential for new insights. To overcome these limitations, in this paper, we propose a novel framework that provides insights into diseases without relying on handcrafted features or human intervention. Our framework is based on deep learning (DL), explainable artificial intelligence (XAI), and clustering. DL is employed to learn deep patterns, enabling efficient differentiation between healthy and pathological images. Explainable artificial intelligence (XAI) visualizes these patterns, and a novel “explanation-weighted” clustering technique is introduced to gain an overview of these patterns across multiple patients. We applied the method to images from the gastrointestinal tract. In addition to real healthy images and real images of polyps, some of the images had synthetic shapes added to represent other types of pathologies than polyps. The results show that our proposed method was capable of organizing the images based on the reasons they were diagnosed as pathological, achieving high cluster quality and a rand index close to or equal to one.

• Khan, Haroon; Qureshi, Nauman Khalid; Yazidi, Anis; Engell, Håvard & Mirtaheri, Peyman (2023). Single-leg stance on a challenging surface can enhance cortical activation in the right hemisphere – A case study. Heliyon. 9(2). doi: 10.1016/j.heliyon.2023.e13628. Full text in Research Archive Show summary

  Maintaining body balance, whether static or dynamic, is critical in performing everyday activities and developing and optimizing basic motor skills. This study investigates how a professional alpine skier’s brain activates on the contralateral side during a single-leg stance. Continuous-wave functional near-infrared spectroscopy (fNIRS) signals were recorded with sixteen sources and detectors over the motor cortex to investigate brain hemodynamics. Three different tasks were performed: barefooted walk (BFW), right-leg stance (RLS), and left-leg stance (LLS). The signal processing pipeline includes channel rejection, the conversation of raw intensities into hemoglobin concentration changes using modified Beer-Lambert law, baseline zero-adjustments, z-normalization, and temporal filtration. The hemodynamic brain signal was estimated using a general linear model with a 2-gamma function. Measured activations (𝑡-values) with 𝑝-value < 0.05 were only considered as statistically significant active channels. Compared to all other conditions, BFW has the lowest brain activation. LLS is associated with more contralateral brain activation than RLS. During LLS, higher brain activation was observed across all brain regions. The right hemisphere has comparatively more activated regions-of-interest. Higher ΔHbO demands in the dorsolateral prefrontal, pre-motor, supplementary motor cortex, and primary motor cortex were observed in the right hemisphere relative to the left which explains higher energy demands for balancing during LLS. Broca’s temporal lobe was also activated during both LLS and RLS. Comparing the results with BFW– which is considered the most realistic walking condition–, it is concluded that higher demands of ΔHbO predict higher motor control demands for balancing. The participant struggled with balance during the LLS, showing higher ΔHbO in both hemispheres compared to two other conditions, which indicates the higher requirement for motor control to maintain balance. A post-physiotherapy exercise program is expected to improve balance during LLS, leading to fewer changes to ΔHbO.

• Glover, Tom Eivind; Lind, Pedro; Yazidi, Anis; Osipov, Evgeny & Nichele, Stefano (2023). Investigating Rules and Parameters of Reservoir Computing with Elementary Cellular Automata, with a Criticism of Rule 90 and the Five-Bit Memory Benchmark. Complex Systems. 32(3), p. 309–351. doi: 10.25088/ComplexSystems.32.3.309. Full text in Research Archive Show summary

  Reservoir computing with cellular automata (ReCAs) is a promising concept by virtue of its potential for effective hardware implementation. In this paper, we explore elementary cellular automata rules in the context of ReCAs and the 5-bit memory benchmark. We combine elementary cellular automaton theory with our results and use them to identify and explain some of the patterns found. Furthermore, we use these findings to expose weaknesses in the 5-bit memory benchmark as it is typically applied in ReCAs, such as pointing out what features it selects for or solving it using random vectors. We look deeply into previously successful rules in ReCAs such as rule 90 and explain some of the consequences of its additive properties as well as the correlation between grid size and performance. Additionally, we present results from exhaustively exploring ReCAs on key parameters such as distractor period, iterations and grid size. The findings of this paper should motivate the ReCAs community to move away from using the 5-bit memory benchmark as it is being applied today.

• Hassan, Ismail; Oommen, John & Yazidi, Anis (2023). Adaptive learning with artificial barriers yielding Nash equilibria in general games. Knowledge engineering review (Print). ISSN 0269-8889. 38. doi: 10.1017/S0269888923000103. Full text in Research Archive Show summary

  Artificial barriers in Learning Automata (LA) is a powerful and yet under-explored concept although it was first proposed in the 1980s. Introducing artificial non-absorbing barriers makes the LA schemes resilient to being trapped in absorbing barriers, a phenomenon which is often referred to as lock in probability leading to an exclusive choice of one action after convergence. Within the field of LA and reinforcement learning in general, there is a sacristy of theoretical works and applications of schemes with artificial barriers. In this paper, we devise a LA with artificial barriers for solving a general form of stochastic bimatrix game. Classical LA systems possess properties of absorbing barriers and they are a powerful tool in game theory and were shown to converge to game’s of Nash equilibrium under limited information. However, the stream of works in LA for solving game theoretical problems can merely solve the case where the Saddle Point of the game exists in a pure strategy and fail to reach mixed Nash equilibrium when no Saddle Point exists for a pure strategy. Furthermore, we provide experimental results that are in line with our theoretical findings.

• Aoueileyine, Mohamed Ould-Elhassen; Bennouri, Hajar; Berqia, Amine; Lind, Pedro; Haugerud, Hårek & Krejcar, Ondrej [Show all 8 contributors for this article] (2023). Quality of Monitoring Optimization in Underwater Sensor Networks through a Multiagent Diversity-Based Gradient Approach. Sensors. 23(8). doi: 10.3390/s23083877. Full text in Research Archive Show summary

  Due to the complex underwater environment, conventional measurement and sensing methods used for land are difficult to apply directly in the underwater environment. Especially for seabed topography, it is impossible to perform long-distance and accurate detection by electromagnetic waves. Therefore, various types of acoustic and even optical sensing devices for underwater applications have been used. Equipped with submersibles, these underwater sensors can detect a wide underwater range accurately. In addition, the development of sensor technology will be modified and optimized according to the needs of ocean exploitation. In this paper, we propose a multiagent approach for optimizing the quality of monitoring (QoM) in underwater sensor networks. Our framework aspires to optimize the QoM by resorting to the machine learning concept of diversity. We devise a multiagent optimization procedure which is able to both reduce the redundancy among the sensor readings and maximize the diversity in a distributed and adaptive manner. The mobile sensor positions are adjusted iteratively using a gradient type of updates. The overall framework is tested through simulations based on realistic environment conditions. The proposed approach is compared to other placement approaches and is found to achieve a higher QoM with a smaller number of sensors.

• Aoueileyine, Mohamed Ould-Elhassen; Allani, Ramzi; Bouallegue, Ridha & Yazidi, Anis (2023). Coverage Strategy for Small-Cell UAV-Based Networks in IoT Environment. Sensors. 23(21). doi: 10.3390/s23218771. Full text in Research Archive Show summary

  In wireless communication, small cells are low-powered cellular base stations that can be used to enhance the coverage and capacity of wireless networks in areas where traditional cell towers may not be practical or cost-effective. Unmanned aerial vehicles (UAVs) can be used to quickly deploy and position small cells in areas that are difficult to access or where traditional infrastructure is not feasible. UAVs are deployed by telecommunication service providers to provide aerial network access in remote rural areas, disaster-affected areas, or massive-attendance events. In this paper, we focus on the scheduling of beaconing periods as an efficient means of energy consumption optimization. The conducted study provides a sub-modular game perspective of the problem and investigates its structural properties. We also provide a learning algorithm that ensures convergence of the considered UAV network to a Nash equilibrium operating point. Finally, we conduct extensive numerical investigations to assist our claims about the energy and data rate efficiency of the strategic beaconing policy (at Nash equilibrium).

• Tveitstøl, Thomas; Tveter, Mats; Teseyra, Ana Silvina Pérez; Hatlestad-Hall, Christoffer; Yazidi, Anis & Hammer, Hugo Lewi [Show all 7 contributors for this article] (2023). Introducing Region Based Pooling for handling a varied number of EEG channels for deep learning models. Frontiers in Neuroinformatics. 17. doi: 10.3389/fninf.2023.1272791. Full text in Research Archive Show summary

  Introduction: A challenge when applying an artificial intelligence (AI) deep learning (DL) approach to novel electroencephalography (EEG) data, is the DL architecture’s lack of adaptability to changing numbers of EEG channels. That is, the number of channels cannot vary neither in the training data, nor upon deployment. Such highly specific hardware constraints put major limitations on the clinical usability and scalability of the DL models. Methods: In this work, we propose a technique for handling such varied numbers of EEG channels by splitting the EEG montages into distinct regions and merge the channels within the same region to a region representation. The solution is termed Region Based Pooling (RBP). The procedure of splitting the montage into regions is performed repeatedly with different region configurations, to minimize potential loss of information. As RBP maps a varied number of EEG channels to a fixed number of region representations, both current and future DL architectures may apply RBP with ease. To demonstrate and evaluate the adequacy of RBP to handle a varied number of EEG channels, sex classification based solely on EEG was used as a test example. The DL models were trained on 129 channels, and tested on 32, 65, and 129-channels versions of the data using the same channel positions scheme. The baselines for comparison were zero-filling the missing channels and applying spherical spline interpolation. The performances were estimated using 5-fold cross validation. Results: For the 32-channel system version, the mean AUC values across the folds were: RBP (93.34%), spherical spline interpolation (93.36%), and zero-filling (76.82%). Similarly, on the 65-channel system version, the performances were: RBP (93.66%), spherical spline interpolation (93.50%), and zero-filling (85.58%). Finally, the 129-channel system version produced the following results: RBP (94.68%), spherical spline interpolation (93.86%), and zero-filling (91.92%). Conclusion: In conclusion, RBP obtained similar results to spherical spline interpolation, and superior results to zero-filling. We encourage further research and development of DL models in the cross-dataset setting, including the use of methods such as RBP and spherical spline interpolation to handle a varied number of EEG channels.

• Djenouri, Youcef; Belhadi, Asma; Yazidi, Anis; Srivastava, Gautam; Chatterjee, Pushpita & Lin, Jerry Chun-Wei (2023). An intelligent collaborative image-sensing system for disease detection. IEEE Sensors Journal. ISSN 1530-437X. 23(2), p. 947–954. doi: 10.1109/JSEN.2022.3202437. Full text in Research Archive Show summary

  With the growth of smart medical devices and applications in smart hospitals, home care facilities, nursing, and the Internet of Medical Things (IoMT) are becoming more ubiquitous. It uses smart medical devices and cloud computing services, and basic Internet of Things (IoT) technology, to detect key body indicators, monitor health situations, and generate multivariate data to provide just-in-time healthcare services. In this article, we present a novel collaborative disease detection system based on IoMT amalgamated with captured image data. The system can be based on intelligent agents, where every agent explores the interaction between different medical data obtained by smart sensor devices using reinforcement learning as well as targets to detect diseases. The agents then collaborate to make a reliable conclusion about the detected diseases. Intensive experiments were conducted using medical data. The results show the importance of using intelligent agents for disease detection in healthcare decision-making. Moreover, collaboration increases the detection rate, with numerical results showing the superiority of the proposed framework compared with baseline solutions for disease detection.

• Karnati, Mohan; Seal, Ayan; Bhattacharjee, Debotosh; Yazidi, Anis & Krejcar, Ondrej (2023). Understanding Deep Learning Techniques for Recognition of Human Emotions Using Facial Expressions: A Comprehensive Survey. IEEE Transactions on Instrumentation and Measurement. ISSN 0018-9456. 72. doi: 10.1109/TIM.2023.3243661. Full text in Research Archive Show summary

  Emotion recognition plays a significant role in cognitive psychology research. However, measuring emotions is a challenging task. Thus, several approaches have been designed for facial expression recognition (FER). Although, the challenges increase further as the data transit from the laboratory-controlled environment to in-the-wild circumstances, nowadays, applications are overwhelmed by a profusion of deep learning (DL) techniques in real-world problems. DL networks have steadily led to a better understanding of low-dimensional discriminative features from high-dimensional complex face patterns for automatic FER. The modern FER systems based on deep neural networks mainly suffer from two problems: overfitting due to the inadequate availability of training data and complications unassociated with the expressions, such as occlusion, posture, illumination, and identity bias. This study aims to provide a comprehensive survey of the significant DL-based methods that have made a notable contribution to the field of FER. Different components of the methods, such as preprocessing, feature extraction, and classification of facial expressions, are described systematically. Moreover, the discussed approaches are analyzed to compare their performance along with their advantages and limitations. Furthermore, different databases relevant to FER are also explored in this study. Essentially, the main aim of this survey is twofold. The former is to discuss the current scenario of FER approaches and the latter is to present some thoughts on the future directions of facial emotion recognition by machines: what are the obstacles and prospects for FER researchers?

• Zhang, Xuanyu; Zhou, Hao; Yu, Ke; Wu, Xiaofei & Yazidi, Anis (2023). Tsetlin Machine for Sentiment Analysis and Spam Review Detection in Chinese. Algorithms. 16(2). doi: 10.3390/a16020093. Full text in Research Archive Show summary

  In Natural Language Processing (NLP), deep-learning neural networks have superior performance but pose transparency and explainability barriers, due to their black box nature, and, thus, there is lack of trustworthiness. On the other hand, classical machine learning techniques are intuitive and easy to understand but often cannot perform satisfactorily. Fortunately, many research studies have recently indicated that the newly introduced model, Tsetlin Machine (TM), has reliable performance and, at the same time, enjoys human-level interpretability by nature, which is a promising approach to trade off effectiveness and interpretability. However, nearly all of the related works so far have concentrated on the English language, while research on other languages is relatively scarce. So, we propose a novel method, based on the TM model, in which the learning process is transparent and easily-understandable for Chinese NLP tasks. Our model can learn semantic information in the Chinese language by clauses. For evaluation, we conducted experiments in two domains, namely sentiment analysis and spam review detection. The experimental results showed thatm for both domains, our method could provide higher accuracy and a higher F1 score than complex, but non-transparent, deep-learning models, such as BERT and ERINE.

• Sharma, Krishna Kumar; Seal, Ayan; Yazidi, Anis & Krejcar, Ondrej (2023). S-Divergence-Based Internal Clustering Validation Index. International Journal of Interactive Multimedia and Artificial Intelligence (IJIMAI). 8(4), p. 127–139. doi: 10.9781/ijimai.2023.10.001. Full text in Research Archive Show summary

  A clustering validation index (CVI) is employed to evaluate an algorithm’s clustering results. Generally, CVI statistics can be split into three classes, namely internal, external, and relative cluster validations. Most of the existing internal CVIs were designed based on compactness (CM) and separation (SM). The distance between cluster centers is calculated by SM, whereas the CM measures the variance of the cluster. However, the SM between groups is not always captured accurately in highly overlapping classes. In this article, we devise a novel internal CVI that can be regarded as a complementary measure to the landscape of available internal CVIs. Initially, a database’s clusters are modeled as a non-parametric density function estimated using kernel density estimation. Then the S-divergence (SD) and S-distance are introduced for measuring the SM and the CM, respectively. The SD is defined based on the concept of Hermitian positive definite matrices applied to density functions. The proposed internal CVI (PM) is the ratio of CM to SM. The PM outperforms the legacy measures presented in the literature on both superficial and realistic databases in various scenarios, according to empirical results from four popular clustering algorithms, including fuzzy k-means, spectral clustering, density peak clustering, and density-based spatial clustering applied to noisy data.

• Fritsch, Lothar; Alzarqawee, Aws Naser Jaber & Yazidi, Anis (2023). An overview of artificial intelligence used in malware. In Zouganeli, Evi; Yazidi, Anis; Mello, Gustavo Borges Moreno e & Lind, Pedro (Ed.), Nordic Artificial Intelligence Research and Development: 4th Symposium of the Norwegian AI Society, NAIS 2022, Oslo, Norway, May 31-June 1, 2022, Revised Selected Papers. Springer Nature. ISSN 9783031170294. Full text in Research Archive Show summary

  Artificial intelligence (AI) and machine learning (ML) methods are increasingly adopted in cyberattacks. AI supports the establishment of covert channels, as well as the obfuscation of malware. Additionally, AI results in new forms of phishing attacks and enables hardto- detect cyber-physical sabotage. Malware creators increasingly deploy AI and ML methods to improve their attack’s capabilities. Defenders must therefore expect unconventional malware with new, sophisticated and changing features and functions. AI’s potential for automation of complex tasks serves as a challenge in the face of defensive deployment of anti-malware AI techniques. This article summarizes the state of the art in AI-enhanced malware and the evasion and attack techniques it uses against AI-supported defensive systems. Our findings include articles describing targeted attacks against AI detection functions, advanced payload obfuscation techniques, evasion of networked communication with AI methods, malware for unsupervised-learning-based cyberphysical sabotage, decentralized botnet control using swarm intelligence and the concealment of malware payloads within neural networks that fulfill other purposes. Keywords: Information Security, Artificial Intelligence, Malware, Steganography, Covert Channels, Machine Learning, Adverse Artificial Intelligence

• Meng, Li; Goodwin, Morten; Yazidi, Anis & Engelstad, Paal E. (2023). Unsupervised State Representation Learning in Partially Observable Atari Games. In Tsapatsoulis, Nicolas (Eds.), Computer Analysis of Images and Patterns. CAIP 2023. Springer. ISSN 9783031442391. p. 212–222. doi: 10.1007/978-3-031-44240-7_21. Full text in Research Archive Show summary

  State representation learning aims to capture latent factors of an environment. Although some researchers realize the connections between masked image modeling and contrastive representation learning, the effort is focused on using masks as an augmentation technique to represent the latent generative factors better. Partially observable environments in reinforcement learning have not yet been carefully studied using unsupervised state representation learning methods. In this article, we create an unsupervised state representation learning scheme for partially observable states. We conducted our experiment on a previous Atari 2600 framework designed to evaluate representation learning models. A contrastive method called Spatiotemporal DeepInfomax (ST-DIM) has shown state-of-the-art performance on this benchmark but remains inferior to its supervised counterpart. Our approach improves ST-DIM when the environment is not fully observable and achieves higher F1 scores and accuracy scores than the supervised learning counterpart. The mean accuracy score averaged over categories of our approach is 66%, compared to 38% of supervised learning. The mean F1 score is 64% to 33%. The code can be found on https://github.com/mengli11235/MST_DIM.

• Karnati, Mohan; Seal, Ayan; Sahu, Geet; Yazidi, Anis & Krejcar, Ondrej (2022). A novel multi-scale based deep convolutional neural network for detecting COVID-19 from X-rays. Applied Soft Computing. ISSN 1568-4946. 125. doi: 10.1016/j.asoc.2022.109109. Full text in Research Archive Show summary

  The COVID-19 pandemic has posed an unprecedented threat to the global public health system, primarily infecting the airway epithelial cells in the respiratory tract. Chest X-ray (CXR) is widely available, faster, and less expensive therefore it is preferred to monitor the lungs for COVID-19 diagnosis over other techniques such as molecular test, antigen test, antibody test, and chest computed tomography (CT). As the pandemic continues to reveal the limitations of our current ecosystems, researchers are coming together to share their knowledge and experience in order to develop new systems to tackle it. In this work, an end-to-end IoT infrastructure is designed and built to diagnose patients remotely in the case of a pandemic, limiting COVID-19 dissemination while also improving measurement science. The proposed framework comprises six steps. In the last step, a model is designed to interpret CXR images and intelligently measure the severity of COVID-19 lung infections using a novel deep neural network (DNN). The proposed DNN employs multi-scale sampling filters to extract reliable and noise-invariant features from a variety of image patches. Experiments are conducted on five publicly available databases, including COVIDx, COVID-19 Radiography, COVIDXRay-5K, COVID-19-CXR, and COVIDchestxray, with classification accuracies of 96.01%, 99.62%, 99.22%, 98.83%, and 100%, and testing times of 0.541, 0.692, 1.28, 0.461, and 0.202 s, respectively. The obtained results show that the proposed model surpasses fourteen baseline techniques. As a result, the newly developed model could be utilized to evaluate treatment efficacy, particularly in remote locations.

• Osnes, Idun; Yazidi, Anis; Jacobsen, Hans-Arno; Eliassen, Frank & Sartori, Sabrina (2022). Harnessing Task Usage Prediction and Latency Sensitivity for Scheduling Workloads in Wind-Powered Data Centers. Energies. 15(12). doi: 10.3390/en15124469. Full text in Research Archive Show summary

  The growing number of data centers consumes a vast amount of energy for processing. There is a desire to reduce the environmental footprint of the IT industry, and one way to achieve this is to use renewable energy sources. A challenge with using renewable resources is that the energy output is irregular as a consequence of the intermittent nature of this form of energy. In this paper, we propose a simple and yet efficient latency-aware workload scheduler that creates an energy-agile workload, by deferring tasks with low latency sensitivity to periods with excess renewable energy. The scheduler also increases the overall efficiency of the data center, by packing the workload into as few servers as possible, using neural-network-based predictions of resource usage on an individual task basis to avoid unnecessarily provisioning an excess number of servers. The scheduler was tested on a subset of real-world workload traces, and real-world wind-power generation data, simulating a small-scale data center co-located with a wind turbine. Extensive experimental results show that the devised scheduler reduced the number of servers doing work in periods of low wind-power production up to 93% of the time, by postponing tasks with a low latency sensitivity to a later interval.

• Abolghasemi, Roza; Engelstad, Paal E.; Herrera-Viedma, Enrique & Yazidi, Anis (2022). A personality-aware group recommendation system based on pairwise preferences. Information Sciences. ISSN 0020-0255. 595(595), p. 1–17. doi: 10.1016/j.ins.2022.02.033. Full text in Research Archive Show summary

  Human personality plays a crucial role in decision-making and it has paramount importance when individuals negotiate with each other to reach a common group decision. Such situations are conceivable, for instance, when a group of individuals want to watch a movie together. It is well known that people influence each other’s decisions, the more assertive a person is, the more influence they will have on the final decision. In order to obtain a more realistic group recommendation system (GRS), we need to accommodate the assertiveness of the different group members’ personalities. Although pairwise preferences are long-established in group decision-making (GDM), they have received very little attention in the recommendation systems community. Driven by the advantages of pairwise preferences on ratings in the recommendation systems domain, we have further pursued this approach in this paper, however we have done so for GRS. We have devised a three-stage approach to GRS in which we 1) resort to three binary matrix factorization methods, 2) develop an influence graph that includes assertiveness and cooperativeness as personality traits, and 3) apply an opinion dynamics model in order to reach consensus. We have shown that the final opinion is related to the stationary distribution of a Markov chain associated with the influence graph. Our experimental results demonstrate that our approach results in high precision and fairness.

• Hassan, Ismail; Oommen, John B. & Yazidi, Anis (2022). Learning Automata with Artificial Reflecting Barriers in Games with Limited Information. Proceedings of the International Florida Artificial Intelligence Research Society Conference. ISSN 2334-0754. 35(35). doi: 10.32473/flairs.v35i.130850. Full text in Research Archive Show summary

  This paper deals with the problem of solving stochastic games (which have numerous business and economic applications), using the interesting tools of Learning Automata (LA), the precursors to Reinforcement Learning (RL). Classical LA systems that possess properties of absorbing barriers, have been used as powerful tools in game theory to devise solutions that converge to the game's Nash equilibrium under limited information(Sastry, Phansalkar, and Thathachar 1994). Games with limited information are intrinsically hard because the player does not know the actions chosen of other players, neither their outcomes. The player might not be even aware of the fact that he/she is playing against an opponent. With the state-of-the-art, the numerous works in LA applicable for solving game theoretical problems, can merely solve the case where the game possesses a Saddle Point in a pure strategy. They are unable to reach mixed Nash equilibria when a Saddle Point is non-existent in pure strategies. Additionally, within the field of LA and RL in general, the theoretical and applied schemes of LA with artificial barriers are scarce, even though incorporating artificial barriers in LA has served as a powerful and yet under-explored concept, since its inception in the 1980’s. More recently, the phenomenon of introducing artificial non-absorbing barriers was pioneered, and this renders the LA schemes to be resilient to being trapped in absorbing barriers. In this paper, we devise a LA with artificial barriers for solving a general form of stochastic bimatrix games. The problem’s complexity has been augmented with the scenario that we consider games in which there is no Saddle Point. By resorting to the above-mentioned powerful concept of artificial reflecting barriers, we propose a LA that converges to an optimal mixed Nash equilibrium even though there may be no Saddle Point when a pure strategy is invoked.

• Seal, Ayan; Bajpai, Rishabh; Karnati, Mohan; Agnihotri, Jagriti; Yazidi, Anis & Herrera-Viedma, Enrique [Show all 7 contributors for this article] (2022). Benchmarks for machine learning in depression discrimination using electroencephalography signals. Applied Intelligence - The International Journal of Research on Intelligent Systems for Real Life Complex Problems. ISSN 0924-669X. doi: 10.1007/s10489-022-04159-y. Full text in Research Archive Show summary

  Diagnosis of depression using electroencephalography (EEG) is an emerging field of study. When mental health facilities are unavailable, the use of EEG as an objective measure for depression management at an individual level becomes necessary. However, the limited availability of the openly accessible EEG datasets for depression and the non-standard task paradigm confine the scope of the research. This study contributes to the area by presenting a dataset that includes EEG data of subjects in the resting state and Patient Health Questionnaire (PHQ)-9 scores. These recordings incorporate EEG signals under both eyes open (EO) and eyes closed (EC) conditions. Moreover, this work documents high performance on various benchmark depression classification tasks with the help of traditional supervised machine learning algorithms, namely Decision Tree, Random Forest, k-Nearest Neighbours, Naive Bayes, Support Vector Machine, Multi-Layer Perceptron, and extreme gradient boosted trees (XGBoost) using the newly created dataset, where the class label of each patient is determined by the PHQ-9 score of the person. Then, feature selection is performed on twenty-three linear, nonlinear, time domain, and frequency domain features using ANOVA test and correlation analysis to identify statistically significant features, which are further fed into algorithms mentioned above separately for distinguishing healthy subjects from depressed. Among these classifiers, the performance of the XGBoost is found to be the best, with an accuracy of 87% for the EO state. The obtained results demonstrate that the proposed method outperforms fourteen existing approaches. The dataset presented in this work can be downloaded via https://drive.google.com/drive/folders/1ANUC-6hq02QG728ZWv2a1UWTLUbRrq y?usp=sharing.

• Caceres, Veronica Alejandra Torres; Duffaut, Kenneth; Yazidi, Anis; Westad, Frank Ove & Johansen, Yngve Bolstad (2022). Automated Well-Log Depth Matching - 1D Convolutional Neural Networks vs. Classic Cross Correlation. Petrophysics. ISSN 1529-9074. 63(1), p. 12–34. doi: 10.30632/PJV63N1-2022a2. Full text in Research Archive Show summary

  During drilling and logging, depth alignment of well logs acquired in the same borehole section at different times is a vital preprocessing step before any petrophysical analysis. Depth alignment requires high precision as depth misalignment between different log curve measurements can substantially suppress possible correlations between formation properties, leading to imprecise interpretation or even misinterpretation. Standard depth alignment involves cross-correlation, which typically requires user intervention for reliability. To improve the depth alignment process, we apply deep learning techniques and propose a simple and practical implementation of a 1-dimensional supervised Convolutional Neural Network (1D CNN). We train seven CNN models using different log measurements such as gamma-ray, resistivity, P- and S-wave sonic, density, neutron, and photoelectric factor (PEF), to estimate depth mismatches between corresponding raw logging while drilling (LWD) and electrical wireline logging (EWL) logs of each measurement type. Our deep learning approach avoids manual feature extraction; hence no high level petrophysical knowledge is needed by our algorithms. We use log data from six wells from the Ivar Aasen Field, in the Norwegian North Sea. Four of the six wells constitute the entire dataset for training and model selection, in which we compare three search algorithms during the hyperparameter tuning. Only two wells have both LWD and EWL log suites. These wells are used for depth shift inference. We focus on estimating bulk shifts, and we assume the existence of small pattern differences. We assess our results by visual inspection, and quantitative metrics such as the Pearson correlation and Euclidean distance. We also compare the CNN depth shifts with depth shifts obtained using the classical cross-correlation method. The CNN performs well and is competitive with cross-correlation. CNN performs better for some log types, resistivity for instance, than others. Several factors influence our results including, the quality of the input data, borehole conditions, pattern differences between LWD and EWL, and significant stretch/squeeze effects. Differences between the mean Pearson correlation computed after CNN and cross-correlation depth matching process are of the order of 10-1 and 10-2. Our CNN approach is therefore a potential alternative to current depth matching methods, which may reduce the amount of user intervention required from the petrophysicist.

• Aryan, Ramtin; Yazidi, Anis; Brattensborg, Frode Henriksen; Kure, Øivind & Engelstad, Paal E. (2022). SDN Spotlight: A real-time OpenFlow troubleshooting framework. Future Generation Computer Systems. ISSN 0167-739X. 133, p. 364–377. doi: 10.1016/j.future.2022.03.014. Full text in Research Archive Show summary

  Troubleshooting in SDN-based networks is still a cumbersome task that can overwhelm human attention. Various anomalies, such as installation failure, disordered rules, and loops, remain unnoticed even when the most recent detection methods are used. In this paper, we address the issue of verifying SDN policies by actively probing the data plane. SDN Spotlight is presented as an anomaly detection framework that tries to detect installation failures, rule conflicts, and loops. In contrast to recent work, such as Monocle and Pronto, SDN Spotlight verifies a chain of rules using a single probing packet. This approach also reduces the number of monitoring rules, which has a direct effect on saving TCAM memory usage and minimizing the packet matching time. SDN Spotlight addresses two problems: verifying rule installation and forwarding behavior verification. Within the SDN Spotlight framework, we introduce two different approaches for forwarding anomaly detection: Hedge-SDN Spotlight and Open-SDN Spotlight. Furthermore, we devise an efficient and fast probe generation algorithm that generates one single probing packet per chain of rules. As opposed to other related work, Hedge-SDN Spotlight does not yield false positives and false negatives when detecting loops and forwarding failures. The results of the experiment demonstrate that SDN Spotlight is much faster than the SDNProbe and SDN traceroute method, in some cases by a factor of up to seven times as fast.

• Karnati, Mohan; Seal, Ayan; Yazidi, Anis & Krejcar, Ondrej (2022). FLEPNet: Feature Level Ensemble Parallel Network for Facial Expression Recognition. IEEE Transactions on Affective Computing. 13(4), p. 2058–2070. doi: 10.1109/TAFFC.2022.3208309. Full text in Research Archive
• Hammer, Hugo Lewi; Yazidi, Anis & Rue, Håvard (2022). Estimating Tukey depth using incremental quantile estimators. Pattern Recognition. ISSN 0031-3203. 122. doi: 10.1016/j.patcog.2021.108339. Full text in Research Archive Show summary

  Measures of distance or how data points are positioned relative to each other are fundamental in pattern recognition. The concept of depth measures how deep an arbitrary point is positioned in a dataset, and is an interesting concept in this regard. However, while this concept has received a lot of attention in the statistical literature, its application within pattern recognition is still limited. To increase the applicability of the depth concept in pattern recognition, we address the well-known computational challenges associated with the depth concept, by suggesting to estimate depth using incremental quantile estimators. The suggested algorithm can not only estimate depth when the dataset is known in advance, but can also track depth for dynamically varying data streams by using recursive updates. The tracking ability of the algorithm was demonstrated based on a real-life application associated with detecting changes in human activity from real-time accelerometer observations. Given the flexibility of the suggested approach, it can detect virtually any kind of changes in the distributional patterns of the observations, and thus outperforms detection approaches based on the Mahalanobis distance.

• Djenouri, Youcef; Belhadi, Asma; Yazidi, Anis; Srivastava, Gautam & Lin, Jerry Chun-Wei (2022). Artificial intelligence of medical things for disease detection using ensemble deep learning and attention mechanism. Expert Systems. ISSN 0266-4720. doi: 10.1111/exsy.13093. Full text in Research Archive Show summary

  In this paper, we present a novel paradigm for disease detection. We build an artificial intelligence based system where various biomedical data are retrieved from distributed and homogeneous sensors. We use different deep learning architectures (VGG16, RESNET, and DenseNet) with ensemble learning and attention mechanisms to study the interactions between different biomedical data to detect and diagnose diseases. We conduct extensive testing on biomedical data. The results show the benefits of using deep learning technologies in the field of artificial intelligence of medical things to diagnose diseases in the healthcare decision-making process. For example, the disease detection rate using the proposed methodology achieves 92%, which is greatly improved compared to the higher-level disease detection models.

• Palumbo, Fabrizio; Buji, Adam; Yazidi, Anis & Haugerud, Hårek (2022). Genetic Algorithms For Tightening Security. In Ukjent, Ukjent (Eds.), Proceedings of the 14th IFIP Wireless and Mobile Networking Conference. IEEE (Institute of Electrical and Electronics Engineers). ISSN 9783903176522. doi: 10.23919/WMNC56391.2022.9954297. Full text in Research Archive Show summary

  Proper configuration of operating systems and program parameters is known to be a key security factor in order to remove vulnerabilities. It is known that vulnerabilities can be caused by a human misconfiguration or by an improper chain of parameter settings. It is impossible to find an optimal combination manually due to the enormous number of possible configurations. In this article, we resort to a Genetic Algorithm equipped with a user-defined fitness function in order to compute a configuration of high fitness. Our work presents a two-fold contribution. First, we successfully use a GA to implement a moving target defense by alerting the configuration regularly in order to spoil an attacker's reconnaissance efforts. The GA tightens the security solution by evolving the fitness of the configuration over generations while maintaining diversity within generations across a pool of servers. This resulted in high-quality configurations crucial for a successful moving target defense strategy. Second, we try to find a compromise between tightening the security of the configuration and maintaining the Quality of Service (QoS) on a web server. In practice, usually tightening security on a web server comes at the cost of a decrease in QoS.

• Djenouri, Youcef; Srivastava, Gautam; Yazidi, Anis & Lin, Jerry Chun-Wei (2022). An edge-driven multi-agent optimization model for infectious disease detection. Applied Intelligence - The International Journal of Research on Intelligent Systems for Real Life Complex Problems. ISSN 0924-669X. 52(12), p. 14362–14373. doi: 10.1007/s10489-021-03145-0. Full text in Research Archive Show summary

  This research work introduces a new intelligent framework for infectious disease detection by exploring various emerging and intelligent paradigms. We propose new deep learning architectures such as entity embedding networks, long-short term memory, and convolution neural networks, for accurately learning heterogeneous medical data in identifying disease infection. The multi-agent system is also consolidated for increasing the autonomy behaviours of the proposed framework, where each agent can easily share the derived learning outputs with the other agents in the system. Furthermore, evolutionary computation algorithms, such as memetic algorithms, and bee swarm optimization controlled the exploration of the hyper-optimization parameter space of the proposed framework. Intensive experimentation has been established on medical data. Strong results obtained confirm the superiority of our framework against the solutions that are state of the art, in both detection rate, and runtime performance, where the detection rate reaches 98% for handling real use cases.

• Haugerud, Hårek; Sobhie, Mohamad & Yazidi, Anis (2022). Tuning of Elasticsearch Configuration: Parameter Optimization Through Simultaneous Perturbation Stochastic Approximation. Frontiers in Big Data. 5. doi: 10.3389/fdata.2022.686416. Full text in Research Archive Show summary

  Elasticsearch is currently the most popular search engine for full-text database management systems. By default, its configuration does not change while it receives data. However, when Elasticsearch stores a large amount of data over time, the default configuration becomes an obstacle to improving performance. In addition, the servers that host Elasticsearch may have limited resources, such as internal memory and CPU. A general solution to these problems is to dynamically tune the configuration parameters of Elasticsearch in order to improve its performance. The sheer number of parameters involved in this configuration makes it a complex task. In this work, we apply the Simultaneous Perturbation Stochastic Approximation method for optimizing Elasticsearch with multiple unknown parameters. Using this algorithm, our implementation optimizes the Elasticsearch configuration parameters by observing the performance and automatically changing the configuration to improve performance. The proposed solution makes it possible to change the configuration parameters of Elasticsearch automatically without having to restart the currently running instance of Elasticsearch. The results show a higher than 40% improvement in the combined data insertion capacity and the system’s response time.

• Caceres, Veronica Alejandra Torres; Duffaut, Kenneth; Yazidi, Anis; Westad, Frank Ove & Johansen, Yngve Bolstad (2022). Automated well log depth matching: Late fusion multimodal deep learning. Geophysical Prospecting. ISSN 0016-8025. p. 1–28. doi: 10.1111/1365-2478.13200. Full text in Research Archive Show summary

  Petrophysical interpretation and optimal correlation extraction of different measurements require accurate well log depth matching. We have developed a supervised multimodal machine learning alternative for the task of simultaneously matching raw logging while drilling and electrical wireline logging logs. Seven one-dimensional convolutional neural networks are trained using different log measurements: gamma-ray, resistivity, P- and S-wave sonic, density, neutron and photoelectric factors, and their depth shift estimates are aggregated using different multimodal late fusion strategies. We test the late fusion average, late fusion weighted average, late fusion with linear and nonlinear learners and model-level fusion. Depth matching results using the different fusion strategies applied to two unseen wells are compared using visual inspection and the mean Pearson correlation. All models perform well, increasing the correlation after depth matching. Late fusion weighted average achieves the highest scores for all log types. The late fusion weighted average results are compared to a cross-correlation user-assisted workflow and manual depth matching for validation. In general, the convolutional neural network fused method exhibits a lower performance than the traditional methods. For one of the wells, the cross-correlation shows higher correlation values than the other methods but for the second well the manual depth match performs best. However, the differences in Pearson correlation values are small ranging from 0.01 to 0.1. The manual depth match performs very well for the sonic logs, which tend to require slightly larger depth shifts than the other measurements, thus a common depth shift might not always be suitable. Although our convolutional neural network fused approach is limited to estimating bulk shifts and uses constant fusion weights, its performance is similar to that of more time-consuming methods. Our approach might be substantially improved by including dynamic shifts (stretch/squeeze) and depth-dependent fusion weights via long-short-term memory recurrent neural networks.

• Nguyen, Thu; Khadka, Rabindra; Phan, Nhan; Yazidi, Anis; Halvorsen, Pål & Hammer, Hugo Lewi [Show all 7 contributors for this article] (2022). Combining datasets to improve model fitting, 2022 International Joint Conference on Neural Networks (IJCNN). IEEE (Institute of Electrical and Electronics Engineers). ISSN 9781728186719. doi: 10.1109/IJCNN54540.2023.10191273. Full text in Research Archive
• Karnati, Mohan; Seal, Ayan; Yazidi, Anis & Krejcar, Ondrej (2022). LieNet: A deep convolution neural Network framework for detecting deception. IEEE Transactions on Cognitive and Developmental Systems. ISSN 2379-8920. 14(3), p. 971–984. doi: 10.1109/TCDS.2021.3086011. Full text in Research Archive Show summary

  Nowadays, automatic deception detection has received considerable attention in the machine learning community owing to this research interest to its vast applications in the fields of social media, interviews, law enforcement, and the military. In this study, a novel deep convolution neural network (DCNN) named LieNet is proposed to precisely detect the multiscale variations of deception automatically. Our approach is a combination of contact and noncontact-based approaches. First, 20 frames from each video are fetched and concatenated to form a single image. Moreover, an audio signal is extracted from video and treated as image input by plotting the signal into 2-D plane. Furthermore, 13 channels of electroencephalogram signals are plotted into 2-D plane and concatenated to generate an image. Second, the LieNet model extracts features from each modality separately. Third, scores are estimated using a softmax classifier for all the modalities. Finally, three scores are combined using score level fusion to obtain a score, which gives support in favor of either deception or truth. The LieNet is validated on the “Bag-of-Lies (BoL),” “ real-life (RL) trail,” and “Miami University Deception Detection (MU3D)” databases by considering four evaluation indexes, viz., accuracy, precision, recall, and F1-score. Experimental outcomes depict that the LieNet defeats an initial work on Set-A and Set-B of the BoL database with average accuracies of 95.91% and 96.04%. respectively. The accuracies obtained by the LieNet are 97% and 98% on RL trail and MU3D databases respectively.

• Fineide, Fredrik; Storås, Andrea; Chen, Xiangjun; Magnø, Morten Schjerven; Yazidi, Anis & Riegler, Michael [Show all 7 contributors for this article] (2022). Predicting an unstable tear film through artificial intelligence. Scientific Reports. 12(1). doi: 10.1038/s41598-022-25821-y. Full text in Research Archive Show summary

  Dry eye disease is one of the most common ophthalmological complaints and is defined by a loss of tear film homeostasis. Establishing a diagnosis can be time-consuming, resource demanding and unpleasant for the patient. In this pilot study, we retrospectively included clinical data from 431 patients with dry eye disease examined in the Norwegian Dry Eye Clinic to evaluate how artificial intelligence algorithms perform on clinical data related to dry eye disease. The data was processed and subjected to numerous machine learning classification algorithms with the aim to predict decreased tear film break-up time. Moreover, feature selection techniques (information gain and information gain ratio) were applied to determine which clinical factors contribute most to an unstable tear film. The applied machine learning algorithms outperformed baseline classifications performed with ZeroR according to included evaluation metrics. Clinical features such as ocular surface staining, meibomian gland expressibility and dropout, blink frequency, osmolarity, meibum quality and symptom score were recognized as important predictors for tear film instability. We identify and discuss potential limitations and pitfalls.

• Abolghasemi, Roza; Khadka, Rabindra; Lind, Pedro; Engelstad, Paal E.; Viedma, Enrique Herrera & Yazidi, Anis (2022). Predicting missing pairwise preferences from similarity features in group decision making. Knowledge-Based Systems. ISSN 0950-7051. 256(256). doi: 10.1016/j.knosys.2022.109860. Full text in Research Archive Show summary

  In group decision-making (GDM), fuzzy preference relations (FPRs) refer to pairwise preferences in the form of a matrix. Within the field of GDM, the problem of estimating missing values is of utmost importance, since many experts provide incomplete preferences. In this paper, we propose a new method called the entropy-based method for estimating the missing values in the FPR. We compared the accuracy of our algorithm for predicting the missing values with the best candidate algorithm from state of the art achievements. In the proposed entropy-based method, we took advantage of pairwise preferences to achieve good results by storing extra information compared to single rating scores, for example, a pairwise comparison of alternatives vs. the alternative’s score from one to five stars. The entropy-based method maps the prediction problem into a matrix factorization problem, and thus the solution for the matrix factorization can be expressed in the form of latent expert features and latent alternative features. Thus, the entropy-based method embeds alternatives and experts in the same latent feature space. By virtue of this embedding, another novelty of our approach is to use the similarity of experts, as well as the similarity between alternatives, to infer the missing values even when only minimal data are available for some alternatives from some experts. Note that current approaches may fail to provide any output in such cases. Apart from estimating missing values, another salient contribution of this paper is to use the proposed entropy-based method to rank the alternatives. It is worth mentioning that ranking alternatives have many possible applications in GDM, especially in group recommendation systems (GRS).

• Omslandseter, Rebekka Olsson; Jiao, Lei; Zhang, Xuan; Yazidi, Anis & Oommen, John (2022). The Hierarchical Discrete Pursuit Learning Automaton: A Novel Scheme with Fast Convergence and Epsilon-Optimality. IEEE Transactions on Neural Networks and Learning Systems. ISSN 2162-237X. 35(6), p. 8278–8292. doi: 10.1109/TNNLS.2022.3226538. Full text in Research Archive Show summary

  Since the early 1960s, the paradigm of learning automata (LA) has experienced abundant interest. Arguably, it has also served as the foundation for the phenomenon and field of reinforcement learning (RL). Over the decades, new concepts and fundamental principles have been introduced to increase the LA’s speed and accuracy. These include using probability updating functions, discretizing the probability space, and using the “Pursuit” concept. Very recently, the concept of incorporating “structure” into the ordering of the LA’s actions has improved both the speed and accuracy of the corresponding hierarchical machines, when the number of actions is large. This has led to the $\epsilon$ -optimal hierarchical continuous pursuit LA (HCPA). This article pioneers the inclusion of all the above-mentioned phenomena into a new single LA, leading to the novel hierarchical discretized pursuit LA (HDPA). Indeed, although the previously proposed HCPA is powerful, its speed has an impediment when any action probability is close to unity, because the updates of the components of the probability vector are correspondingly smaller when any action probability becomes closer to unity. We propose here, the novel HDPA, where we infuse the phenomenon of discretization into the action probability vector’s updating functionality, and which is invoked recursively at every stage of the machine’s hierarchical structure. This discretized functionality does not possess the same impediment, because discretization prohibits it. We demonstrate the HDPA’s robustness and validity by formally proving the $\epsilon$ -optimality by utilizing the moderation property. We also invoke the submartingale characteristic at every level, to prove that the action probability of the optimal action converges to unity as time goes to infinity. Apart from the new machine being $\epsilon$ -optimal, the numerical results demonstrate that the number of iterations required for convergence is significantly reduced for the HDPA, when compared to the state-of-the-art HCPA scheme.

• Maratova, Assem; Lencastre, Pedro; Yazidi, Anis & Lind, Pedro (2022). Comparative Analysis of Functional Connectivity Metrics in EEG Datasets. IEEE Signal Processing in Medicine and Biology Symposium (SPMB). ISSN 2372-7241. doi: 10.1109/SPMB55497.2022.10014890. Full text in Research Archive Show summary

  Analysis of functional connectivity helps to determine how brain regions interact with one another and to understand neurological diseases better. In this study, we compare functional connectivity networks derived from electroencephalogram (EEG) data using Pearson's correlation and mutual information. The TUH EEG Epilepsy Corpus (TUEP) dataset is analysed with methods from Graph Theory, Statistics and Machine Learning. Our findings can be used to develop features for predictive models. Specifically, we show that with just 19 channels, a convolutional neural network model achieves 94% and 95% area under the receiver operating characteristic (ROC) curve (AUC) for correlation and mutual information, respectively. Thus, we provide evidence that application of Machine Learning methods to EEG data not containing seizures can help to accurately identify individuals with epilepsy. This may have considerable implications on diagnosing the pathology.

View all works in NVA

• Brodtkorb, André; Begnum, Kyrre Matthias; Chen, Weiqin; Erikson, Håkon; Fritsch, Lothar & Habib, Laurence [Show all 21 contributors for this article] (2024). Fra EDB til KI. 40 år med informasjonsteknologi. Institutt for informasjonsteknologi, OsloMet. ISBN 9788283646382. 160 p. Full text in Research Archive
• Zouganeli, Evi; Yazidi, Anis; Mello, Gustavo Borges Moreno e & Lind, Pedro (2023). Nordic Artificial Intelligence Research and Development: 4th Symposium of the Norwegian AI Society, NAIS 2022, Oslo, Norway, May 31-June 1, 2022, Revised Selected Papers. Springer Nature. ISBN 9783031170294. 130 p. Full text in Research Archive Show summary

  This book constitutes the refereed proceedings of the 4th Symposium of the Norwegian AI Society, NAIS 2022, held in Oslo, Norway, during May 31-June 1, 2022. The 11 full papers included in this book were carefully reviewed and selected from 17 submissions. They were organized in topical sections as follows: robotics and intelligent systems; ai in cyber and digital sphere; ai in biological applications and medicine; and towards new ai methods. Robotics and Intelligent Systems.- Knowledge Infused Representations through Combination of Expert Knowledge and Original Input.- Cognitive Robotics.- towards the development of next-generation robotics and intelligent systems.- Pattern Based Software Architecture for Predictive Maintenance.- AI in Cyber and Digital Sphere.- An overview of artificial intelligence used in malware.- Fake News Detection by Weakly Supervised Learning Based on Content-Features.- Improving the Usability of Tabular Data through Data Annotation, Repair and Augmentation.- AI in Biological Applications and Medicine.- Detecting human embryo cleavage stages using YOLO v5 object detection algorithm.- Phenotyping of Cervical Cancer Risk Groups via Generalized Low-Rank Models using Medical Questionnaires.- Automatic unsupervised clustering of videos of the intracytoplasmic sperm injection (ICSI) procedure.- Towards New AI Methods.- The Kernelized Taylor Diagram.- Simulating University Application Data for Fair Matchings.

View all works in NVA

• Tran, Thi Thuy Chau; Erdem, Cagri; Yazidi, Anis; Kjølerbakken, Kai Morgan & Griwodz, Carsten (2026). Research Proposal: Non-intrusive Stress Recognition using Multimodality Deep Learning. doi: 10.1145/3793853.3798410. Full text in Research Archive
• Szorkovszky, Alexander; Mathema, Rujeena; Silva, Pedro Rego Lencastre e; Yazidi, Anis & Lind, Pedro (2025). Saccade crossing avoidance as a visual search strategy. Full text in Research Archive
• Acharya, Sushil; Fabian, Karl & Yazidi, Anis (2025). Automated Depth Alignment of Well Logs Using Siamese Neural Networks. doi: 10.30632/spwla-2025-0048. Full text in Research Archive Show summary

  Accurate depth alignment of well logs is essential for re-liable subsurface interpretation, influencing petrophysi-cal analysis. Traditional methods like cross-correlation struggle with repetitive lithologies, often leading to false matches and requiring extensive manual corrections(Torres Caceres et al., 2022). While machine learning offers alternatives, supervised approaches demand large labeled datasets, and unsupervised clustering requires extensive post-processing, introducing potential errors(Acharya and Fabian, 2024b; Acharya et al., 2025). This study proposes a Siamese Neural Network (SNN) to au-tomate log alignment by learning geological similarity between reference and unsynchronized logs. The SNN eliminates the need for labeled data, minimizes post-processing, and effectively handles irregular depth er-rors. Tested on gamma-ray (GR) logs from different logging methods within the same well, the proposed method improves Pearson correlation from 0.003 to 0.878 and reduces Euclidean distance by 74.623%, significantly outperforming cross-correlation in non-linear alignment scenarios. A novel similarity score enhances geological plausibility, complementing statistical metrics and pro-viding a valuable validation tool, especially in complex intervals. While the SNN is computationally slower, its superior accuracy makes it a reliable choice for post-drilling workflows, ensuring more precise depth align-ment with minimal manual intervention. By propagat-ing corrections to co-acquired logs, it enhances multi-log consistency for reservoir characterization. This work provides a scalable and reliable solution to depth alignment challenges, with potential applications in other geoscientific fields requiring log matching.

• Szczepanski, Tomasz Szymon; Omland, Petter Moe; Dunker, Øystein; Sand, Trond Halfdan; Tannemaat, Martijn & Reijntjes, Robert [Show all 11 contributors for this article] (2025). Deriving reference limits from historical data - A comparison of four novel methods. Full text in Research Archive
• Yazidi, Anis (2025). Robust RL with Learning Automata: Structure, Timescales, and Mixed Strategies. Full text in Research Archive
• Radwan, Mohamed; Lind, Pedro & Yazidi, Anis (2024). An Interpretable Graph Based Model for Classification Of EEG using Directional Functional Connectivity. Full text in Research Archive
• Khadka, Rabindra; Lind, Pedro; Mello, Gustavo Borges Moreno E; Riegler, Michael Alexander & Yazidi, Anis (2024). Inducing Inductive Bias in Vision Transformer for EEG Classification. Full text in Research Archive
• Kasahara, Yuri; Schroeder, Daniel Thilo; Lind, Pedro & Yazidi, Anis (2023). Mapping the evolution of anti-Muslim hate speech in Norwegianl social media. Full text in Research Archive
• Hammarstedt, Mark; Løland, Anders; Yazidi, Anis; Zouganeli, Evi; Ugland, Yngvar & Andersen, Eirik [Show all 7 contributors for this article] (2023). How can Norway become a Leading AI Nation, as Sweden & Finland - Panel discussion. Full text in Research Archive
• Holter, Trym; Sunnanmark, Anna-Maija; Yazidi, Anis; Zouganeli, Evi; Pettersen, Klas Henning & Ugland, Yngvar [Show all 7 contributors for this article] (2023). Norway’s Potential and Advantages for AI - Panel discussion. Full text in Research Archive
• Djenouri, Youcef; Belbachir, Ahmed Nabil; Tomasz, Michalak & Yazidi, Anis (2023). Shapley Deep Learning: A Consensus for General-Purpose Vision Systems. Full text in Research Archive
• Haraldsen, Ira Hebold; Hatlestad-Hall, Christoffer; Marra, Camillo; Renvall, Hanna; Maestú, Fernando & Acosta-Hernández, Jorge [Show all 69 contributors for this article] (2023). Intelligent digital tools for screening of brain connectivity and dementia risk estimation in people affected by mild cognitive impairment: the AI-Mind clinical study protocol. Frontiers in Neurorobotics. 17. doi: 10.3389/fnbot.2023.1289406. Full text in Research Archive Show summary

  More than 10 million Europeans show signs of mild cognitive impairment (MCI), a transitional stage between normal brain aging and dementia stage memory disorder. The path MCI takes can be divergent; while some maintain stability or even revert to cognitive norms, alarmingly, up to half of the cases progress to dementia within 5 years. Current diagnostic practice lacks the necessary screening tools to identify those at risk of progression. The European patient experience often involves a long journey from the initial signs of MCI to the eventual diagnosis of dementia. The trajectory is far from ideal. Here, we introduce the AI-Mind project, a pioneering initiative with an innovative approach to early risk assessment through the implementation of advanced artificial intelligence (AI) on multimodal data. The cutting-edge AI-based tools developed in the project aim not only to accelerate the diagnostic process but also to deliver highly accurate predictions regarding an individual’s risk of developing dementia when prevention and intervention may still be possible. AI-Mind is a European Research and Innovation Action (RIA H2020-SC1-BHC-06-2020, No. 964220) financed between 2021 and 2026. First, the AI-Mind Connector identifies dysfunctional brain networks based on high-density magneto- and electroencephalography (M/EEG) recordings. Second, the AI-Mind Predictor predicts dementia risk using data from the Connector, enriched with computerized cognitive tests, genetic and protein biomarkers, as well as sociodemographic and clinical variables. AIMind is integrated within a network of major European initiatives, including The Virtual Brain, The Virtual Epileptic Patient, and EBRAINS AISBL service for sensitive data, HealthDataCloud, where big patient data are generated for advancing digital and virtual twin technology development. AI-Mind’s innovation lies not only in its early prediction of dementia risk, but it also enables a virtual laboratory scenario for hypothesis-driven personalized intervention research. This article introduces the background of the AI-Mind project and its clinical study protocol, setting the stage for future scientific contribution.

• Kasahara, Yuri; Nilsen, Anne Birgitta; Yazidi, Anis & Liodden, Tone (2022). Hatefulle ytringer mot muslimer på norske sosiale medier. Full text in Research Archive
• Acharya, Sushil; Fabian, Karl & Yazidi, Anis (2026). Dynamic Depth Alignment of Wellbore Measurements Using Machine Learning. Norges teknisk-naturvitenskapelige universitet. ISSN 9788235302786. Full text in Research Archive Show summary

  Well logs record physical properties of rock formations along a borehole as a function of depth. They are essential for evaluating subsurface formation properties in hydrocarbon exploration, reservoir development, groundwater studies, and numerous other geoscientific and engineering applications. Despite their importance, well logs acquired by different logging tools, such as electrical wireline logging (EWL) and logging while drilling (LWD), or even by repeat runs of the same tool, often assign the same geological features to slightly different depths because of mechanical effects, operational conditions, and differences in data-processing workflows. For meaningful analysis, measurements from multiple tools or runs must be integrated and interpreted together. Because each run records depth independently, the resulting logs are not automatically aligned to a common depth. The process of aligning these measurements to a single, consistent depth reference is known as well-log depth matching and remains a major challenge in subsurface characterization. In the oil and gas industry, depth matching is performed manually by experts using tie points and correlation-based methods, sometimes supported by dynamic time warping (DTW). Although effective in many cases, these approaches are time-consuming when applied to large datasets, subjective, and less reliable in intervals with repetitive log patterns or non-uniform depth errors. This thesis develops automated and semi-automated machine learning (ML) and deterministic methods for aligning LWD logs to the EWL depth reference within the same well. The aim is to reduce manual effort and provide geologically reliable results that experts can trust and review. Rather than estimating the exact true along-hole depth, the thesis focuses on aligning LWD and EWL logs so that corresponding geological features appear at the same depth. The proposed methods are designed to handle nonuniform depth mismatches, preserve geological consistency, and integrate with practical depth-alignment workflows. The first contribution is a learning-based global alignment framework, the Deep Hierarchical Graph Correlation (DHGC). DHGC learns window-level representations from pairs of gamma ray (GR) logs and uses them to construct a pairwise similarity matrix. A constrained graph-based algorithm, implemented using Dijkstra’s algorithm, then computes a globally consistent nonlinear alignment path. By preserving monotonicity and controlling slope and step size, the method reduces the one-to-many mappings that often arise in ambiguous intervals with standard DTW. The second contribution is a supervised local alignment model called the Attention-Driven Convolutional Neural Network (AD-CNN). The model predicts depth shifts at the window level. Any extreme or unrealistic shifts are corrected using a separate outlier-detection workflow. After outlier correction, the predicted shifts are used either to shift local log segments or to build a continuous depth correction across the well. The thesis also proposes a semi-automated workflow that focuses on geological boundaries to improve interpretability and practical applicability. Unlike the single-log alignment methods described earlier, this workflow performs alignment jointly using multiple logs. The method enhances step-like lithological transitions using the Fast Walsh-Hadamard Transform (FWHT). Low-pass filtering in the Walsh domain highlights step-like structures that reflect formation layering. From this, a Walsh Boundary Index is computed to identify likely formation boundaries. These detected boundaries are then used as control points to build a monotonic, piecewise-linear depth-warping function. This approach is similar to how practitioners manually tie key points between logs. It produces clear intermediate outputs, such as boundary evidence, selected anchor points, and the resulting depth mapping. The method is useful when depth misalignment varies with depth and cannot be corrected by a single global shift. Building on the same idea of boundary-guided alignment, but using data-driven methods instead of deterministic signal processing, the thesis also explores unsupervised multi-log alignment methods. By combining dimensionality-reduction techniques such as PCA or autoencoders with clustering methods such as K-means, the approach identifies clusters that may represent distinct lithological units. It also identifies the boundaries between these units, which may correspond to bed boundaries. This is achieved without requiring explicit depth-shift labels. The proposed methods are evaluated on well-log datasets from the Norwegian Continental Shelf (NCS), using well-wise hold-out validation and comparisons with established industrial baselines, including correlation-based methods and DTW. Performance is evaluated quantitatively using Pearson correlation, Euclidean distance, Earth Mover’s Distance (for DHGC), and runtime, and qualitatively through visual inspection. For alignment based on a single log type, DHGC provides robust global alignment and is faster than DTW. The AD-CNN provides efficient local alignment with accuracy comparable to established methods. For multiple logs, the FWHT-based workflow improves alignment quality while maintaining interpretability. The unsupervised multi-log methods also improve alignment quality across several log types, with the autoencoder combined with K-means approach performing best among the tested unsupervised methods. These results suggest a practical stepwise workflow for industry use: first, apply an initial global shift; then use anchor-point warping where clear markers are available; and finally apply flexible nonlinear methods only in complex intervals. All depth corrections and processing decisions should be stored as metadata, uncertain intervals should be flagged with a “no-align” option, and final verification should use repeat sections or known markers. This work reduces manual effort, improves consistency, and provides clear results for reliable automated depth matching in large-scale subsurface studies.

• Filho, Sidney Pontes; Nichele, Stefano; Yazidi, Anis; Zhang, Jianhua; Hammer, Hugo Lewi & Sandvig, Ioanna [Show all 7 contributors for this article] (2023). Optimization of dynamical systems towards criticality and intelligent behavior. Norges teknisk-naturvitenskapelige universitet. ISSN 9788232669530. Full text in Research Archive Show summary

  With the progress of computing power, artificial intelligence (AI) systems are able to achieve outstanding results that surpass human-level performance on some tasks, such as image recognition. Mainstream AI systems are only able to learn one or more specific tasks, and they commonly fail to go beyond what they were trained for. To have a real breakthrough in the field of AI, those systems require adaptability and the capacity to learn general tasks and data distributions. Such characteristics are among the objectives of artificial general intelligence (AGI) research. The inspiration from biology, neuroscience, and complex systems may guide the development of AGI because AI systems are still rigid instead of self-organizing. Therefore, the research reported in this thesis aims at intelligent dynamical systems that resemble the brain, such as the brain’s critical behavior that may allow the cortex to self-organize to criticality in order to increase computational capacity. The plan to accomplishsuch systems encompasses the usage of optimization methods to find adequate interactions of the components in a complex system, how they are connected with each other, and how they adapt those interactions and connections overtime. The dynamical systems investigated in this research are cellular automata, Boolean networks, and recurrent neural networks with abstract neuron models or with more biologically plausible ones (spiking neurons). Themain optimization method applied in these systems is evolutionary computa-tion or artificial evolution. There are three parts in the presented research. Thefirst uses a deep neural network library to evolve dynamical systems towardscriticality. The second part works on the complexification of spiking neuralnetworks with adaptive synapses for solving tasks in mutable environments.The last one involves the application of neural cellular automata for control-ling robots and even developing their morphology with artificial embryogeny.The results of this research attempt to address some unanswered questionsrelated to the practicality, methodology, and benefits of applying dynamicalsystems in AI.

View all works in NVA