Lectures & learning
Teaching
Recorded lectures and course materials on machine learning, mathematics, and optimization.
Video recordings of online lectures (given during COVID)
Browse by topic. The recordings open on YouTube.
Low-rank optimization
Structured prediction
Distances between distributions based on moments
In RussianGeneral MMD
In Russian
Course materials
Spring 2019Data Analysis and Statistical Learning
Introduction
Statistical learning theory formulation of supervised learning
Linear regression
Logistic regression
Separating hyperplanes (SVM)
Separating hyperplanes (SVM)
Model assessment
Model selection
Tikhonov regularization
l1-regularization (LASSO)
Kernel trick
Midterm
Kernel trick
Feedforward neural networks
Feedforward neural networks
Dimensionality reduction
Dimensionality reduction
Concentration inequalities
Concentration inequalities
Optimization Methods and Techniques
- Nonlinear programming: maxima and minima; first- and second-order conditions.
- Numerical methods for unconstrained optimization: Newton and quasi-Newton methods.
- Constrained optimization: optimality conditions for equality and inequality constraints.
- Gradient projection, Lagrangian relaxation, and penalty methods.
- Linear programming: polyhedra, extreme points, vertices, and basic feasible solutions.
- The simplex method, sensitivity analysis, and strong and weak duality.
- Network flows, the network simplex algorithm, maximum flows, assignment, and shortest paths.
- Integer programming: exact and approximate methods, relaxation, and branch and bound.
- Multi-objective optimization: partial orders, efficiency, scalarization, and applications.
- Goal programming: principal results and applications.