MAP670R-2022: Advanced topics in Deep Learning
Instructor: Edouard Oyallon (edouard.oyallon[at]cnrs[dot]fr)
Notes for the class can be found here
(Chapters 1, 2, 3, 4, 5, 6).
A suggestion of research papers for the
graded projects can be found
here. The first homework can be found here.
Homework and project topics are due on Friday 10th of March. You have to work by pair. Please send your homework by email, with the project you selected and the name of your binome.
(24/02/2023, 14:00-17:30, Amphi Painlevé) Lecture 1: Symmetry,
Invariance & Groups
Material: Chapter 1, 2 of the notes. Slides
Reference papers:
Analyse Fonctionelle, by Laure Saint-Raymond.
A Wavelet Tour of Signal Processing, by Stéphane Mallat.
Understanding Deep
Convolutional Networks, by Stéphane Mallat.
Invariant Scattering
Convolutions Networks, by Joan Bruna, Stéphane Mallat.
(02/03/2023, 14:00-17:30, Amphi Carnot) Lecture 2: Stability to diffeomorphisms and translations.
Material: Chapter 1, 2 of the notes. Slides
Lab: notebook and the digit
Reference paper:
Invariant Scattering Convolutions Networks, by Joan Bruna, Stéphane Mallat.
(10/03/2023, 14:00-17:30, Amphi Painlevé) Lecture 3: The Theory of the Scattering Transform.
Material: Chapter 1, 2 of the notes. Slides
Graded Lab: (/!\ more instructions given during the lecture) notebook
Reference paper:
Group Invariant Scattering, by Stéphane Mallat.
(17/03/2023, 14:00-17:30, Amphi Painlevé) Lecture 4: Approximations results via Shallow Neural Networks
Exercises: pdf
Material: Chapter 3, 4 of the notes. Slides
Reference papers:
Breaking the Curse of Dimensionality with Convex Neural Networks, by Francis Bach.
