Term II 2017-2019
Part I: machine learning basics; deep feedforward networks; convolutional networks; advanced network design;
Part II: approximation theory of deep neural networks; stochastic optimization methods; regularization for deep learning; generalization error of deep neural networks;
Part III: sparse and structured computation; sequence modeling: recurrent and recursive nets; deep reinforcement learning; deep generative models; distributed and decentralized learning.
Term II 2018-2019
This module is designed for graduate students in mathematics. It focuses on modelling problems in real life and other disciplines into mathematical problems and simulating their solutions by scientific computing methods. Major topics covered include modelling and numerical simulations in selected areas of physical and engineering sciences, biology, finance, imaging and optimisation.
Term I 2018-2019
Modeling examples and basic concepts of optimization; convex functions and properties; gradients and subgradients; gradient methods; sub-gradient methods; Newton-type algorithms and the Barzilai-Borwein method; constrained optimization; accelerated proximal gradient methods; stochastic block coordinate descent methods; convex conjugacy and duality; splitting algorithms and implementations; CVX Matlab software for convex programming.
Term I 2017-2018
QR factorization, singular value decomposition, condition numbers, stability, perturbation analysis, least squares problems, eigen value problems.
Spring 2017
Fourier and wavelet analysis, differential and integral calculus in R n , low-dimensional manifolds, inverse and implicit function theorems.
Fall 2016
Direct and iterative solvers for dense and sparse linear systems, QR factorization, eigen decomposition, sparse matrix factorizations, basic parallel computation.
Autumn 2015 and Spring 2016