Optimization and Machine Learning Seminar

About

We are happy to announce a seminar on Optimization and Machine Learning Seminar organized by Corelab at NTUA. The goal of this seminar is to introduce the basic concepts of machine learning and convex optimization.

Automated Learning, or Machine Learning as it is usually called, describes the development of algorithmic procedures that convert experience to expertise. Usually, machine learning procedures are inspired by algorithms used in convex optimization.

In convex optimization we explore algorithmic procedures to optimize convex functions over continous convex domains. Apart from the applications of convex optimization to machine learning, during the last decade there has been a great amount of research on using convex optimization for solving combinatorial optimization problems.

The program of the seminar is described in detail below and involves: (1) the basic definitions and concepts of theoretical machine learning, (2) introduction to convex optimization and basic applications to solving combinatorial problems, (3) the basic definitions, techniques and open problems of deep learning.

Prerequisites - Registration

If you would like to participate in our seminar please register here.

We will assume that all participants are familiar with the following concepts:

  1. Basic Linear Algebra and Probability. A concise and self-contained introduction can be found in Chapters 2, 3 and 4 of this book: Deep Learning.
  2. Lecture Notes on Concentration Inequalities by Michael Goemans: here.

Organizers

Program

Date
Topic
Material
Dec 11
12:30-14:30
Introduction to Learning Theory
Dec 12
15:00-17:00
The Fundamental Theorem of PAC Learning
Dec 14
17:00-19:00
Introduction to Convex Optimization
Dec 18
12:30-14:30
Convex Optimization and Max Flow
Dec 19
15:00-17:00
Online Learning and Mirror Descent
Dec 21
17:00-19:00
Algorithms for Linear Programming
Jan 7
17:00-19:30
Mini-Workshop on Machine Learning

Alex Dimakis Deep Generative Models and Inverse Problems

Christos Tzamos Learning From Positive Εxamples

Constantinos Daskalakis Improving Generative Adversarial Networks using Game Theory and Statistics
Jan 8
12:30-14:30
Linear and Logistic Regression & Introduction to Neural Nets
Jan 9
15:00-17:00
Optimization and Generalization in Neural Nets
& Adversarial Learning
Jan 11
17:00-19:00
Hackathon with PyTorch
Jan 15
12:30-14:30
Generative Models - Non-Convex MLE
Jan 16
15:00-17:00
Rademacher Complexity & Reinforcement Learning

Venue

Room 004 at School of Electrical and Computer Engineering [Directions].