Vegard Antun

Academic interests

Inverse problems, compressive sensing, wavelets, deep learning, neural networks.

For more information see my webpage.

Publications

• V. Antun, F. Renna, C. Poon, B. Adcock, A. C. Hansen. On instabilities of deep learning in image reconstruction and the potential costs of AI. Proc. Natl. Acad. Sci. USA, 2020. [arXiv]. In the news: [The Register] [Cam. Univ. News] [dotmed].

Preprints

• N. Gottschling, V. Antun, B. Adcock, A. C. Hansen. The troublesome kernel: why deep learning for inverse problems is typically unstable, [arXiv].
• V. Antun, Ø. Ryan. On the unification of schemes for wavelets on the interval [Preprint] [Technical report]
• L. Thesing, V. Antun, A. C. Hansen. What do AI algorithms actually learn? - On false structures in deep learning, [arXiv].
• B. Adcock, V. Antun, A. C. Hansen. Uniform recovery in infinite-dimensional compressed sensing and applications to structured binary sampling, [arXiv].

Refereed Conference Articles

• M. Colbrook, V. Antun, A. C. Hansen. On the existence of stable and accurate neural networks for image reconstruction, Signal Processing with Adaptive Sparse Structured
  Representations (SPARS), 2019, [PDF]. 

Slides from selected talks

- How intelligent is artificial intelligence ? - On the surprising and mysterious secrets of deep learning (23. May 2019) link.

- A Practical Guide to Deep Learning at the Department of Mathematics (March 2019) slides, code.
- Trends in mathematics of information (16. May 2018) link.

Courses taught

• Autumn 2019: Lecturing MAT-INF 1100 – Modelling and computations
• Autumn 2016-1018: Plenary exercises in MAT-INF 1100 – Modelling and computations
• Spring 2017: Teaching assistant in MAT-INF 2360 - Applications of linear algebra