Aristide Baratin

Aristide Baratin

PhD Candidate

Mila, Université de Montréal

About me

I’m a mathematical physicist working on artificial intelligence. Currently a PhD candidate at Mila - Quebec AI Institute advised by Simon Lacoste-Julien, I work on foundational aspects of deep learning algorithms, with a focus on understanding their robustness and generalization performance.

Before joining Mila, I was a Humboldt Fellow hosted by the University of Waterloo, and prior to that, a junior scientist at the Max Planck Institute for Gravitational Physics in Potsdam, Germany. My research work was mainly about quantum gravity, and associated mathematical structures from higher dimensional algebra and matrix models. I received a PhD in theoretical physics from Ecole Normale Supérieure de Lyon, under the supervision of Laurent Freidel.

I also love teaching. I’ve taught numerous classes in mathematics, physics and computer science as a course lecturer at UW and McGill, and as a teacher’s assistant at ENS Lyon and UdeM.

See my curriculum vitae for more details.

Interests
  • Theories of deep learning
  • Statistical learning theory
  • Representation learning
  • Applications in computer vision, NLP
Education
  • PhD in Machine Learning, 2017-

    Mila, Université de Montréal

  • PhD in Theoretical Physics, 2009

    ENS Lyon and Perimeter Institute, Waterloo

  • Master's degrees in Mathematics and Physics, 2004

    University Paris-Saclay and ENS Paris

  • Stipendary student, 2002-2004

    ENS Paris-Saclay, Mathematics Department

Publications

(2021). On the Regularity of Attention. arXiv, 2021.

arXiv

(2020). Implicit Regularization via Neural Feature Alignment. AISTATS 2021.

Code Video arXiv

(2020). A Mathematical Theory of Attention. arXiv, 2020.

arXiv

(2019). On the Spectral Bias of Neural Networks. ICML 2019.

Code Slides arXiv

(2018). A Modern Take on the Bias-Variance Tradeoff in Neural Networks. arXiv, 2018.

arXiv

(2018). MINE: Mutual Information Neural Estimation. ICML 2018.

Video arXiv

(2018). A3T: Adversarially Augmented Adversarial Training. NIPS 2017, Machine Deception Workshop.

arXiv

(2017). Exploring Machine Learning for Particle Physics. Technical report, 2017.

PDF Code

Contact