Due to scientific and paternity leaves, this time I will be able to only supervise short master theses that will take place in Spring 2024. Here below two possible projects:
- Explore the use of Dividing Local Gaussian Processes in the chemical space surrounding Vaska's complex. A previous study (Friederich et al, 2020) has shown the excellent predicting performance of Gaussian Process Regression (GPR) in the context of catalysis. Unfortunately, GPR can only be applied to small quantities of data due to mathematical reasons. In this project, the candidate will explore the use of a version of GPR, called Dividing Local Gaussian Processes, that can handle huge amount of data.
- Compare statistical learning methods on a large dataset. Friederich et al (2020) compared a few statistical learning approaches in a very limited chemistry space. The goal of this project is to extend the study to a larger dataset, to evaluate the performance of the statistical learning methods in more challenging situations. A new dataset, tmQMg, has been recently made available (Balcells and Skjelstad, 2020) and can be directly used for the purpose.
Bibliography:
- Balcells, D., & Skjelstad, B. B. (2020). tmQM Dataset—quantum geometries and properties of 86k transition metal complexes. Journal of Chemical Information and Modeling, 60(12), 6135-6146.
- Friederich, P., dos Passos Gomes, G., De Bin, R., Aspuru-Guzik, A., & Balcells, D. (2020). Machine learning dihydrogen activation in the chemical space surrounding Vaska's complex. Chemical Science, 11(18), 4584-4601.