Sylabus

Kurz je rozdelený na 10 trojhodinových blokov. Všetky bloky až na posledné dva budú rozdelené na dve hodiny prednášky a hodinu cvičenia. Témy jednotlivých blokov budú nasledovné:

Blok 1

Motivation for Machine Learning, Introduction to Statistical Learning Theory, Formal Learning Model, Empirical Risk Minimization

Blok 2

PAC Learning, Learning Infinite Hypothesis, No-Free-Lunch Theorem, VC dimension

Blok 3

Machine Learning Prediction Models, Convex Learning Problems

Blok 4

Regularized Risk Minimization, Optimization Essentials, Gradient Descent, Stochastic Gradient Descent

Blok 5

Randomized Coordinate Descent with Arbitrary Sampling

Blok 6

Arbitrary sampling for Empirical Risk Minimization, Minibatching

Blok 7

Accelerated Coordinate Descent with Arbitrary Sampling

Blok 8

Variance Reduced Stochastic Gradient Descent with Arbitrary Sampling

Blok 9

Combutational Lab for 3 hours, experimenting with everything

Blok 10

Challenges for future, Distributed methods, Second-order methods, Non-Convex Loss functions, 3rd hour: Open session for questions