Theoretical Deep Learning
Lecture notes
A brief introduction to supervised learning
Concentration inequalities
SubGaussian, Chernoff bound, Hoeffding's inequality, McDiarmid's inequalty
Uniform bounds and empirical processes
Rademacher complexity, Covering number, Dudley entropy integral
Kernel methods, representer theorem and RKHSs
RKHS II
Twolayer neural networks and the Fourier analysis
The Barron space
Deep neural networks
lecture note,
Approximation theory of deep ResNets
Depth separations
Training neural networks: Convergence?
A brief overview of GD convergence
Slide
Training neural networks beyond the kernel regime
Slide
References
