Introduction to Machine Learning Summer School

June 18, 2018 - June 29, 2018, Chicago, USA

http://www.ttic.edu http://cam.uchicago.edu https://www.nsf.gov/
chicago

 

The Introduction to Machine Learning Summer School is a two week program jointly organized by Toyota Technological Institute at Chicago (TTI Chicago) and the Committee on Computational and Applied Mathematics (CCAM) at Univeristy of Chicago as part of an NSF funded Research Training Group (RTG) grant.

This intensive two week summer school will follow the format of a short course on beginner level introduction to machine learning (see tentative schedule for specific topics). The program is aimed at advanced undergraduate and master students in computer science, mathematics, statistics, and related fields, as well as graduate students from other disciplines who would like to better understand and use machine learning.
As a result of participating in this program, a student is expected to

  1. understand the goals and capabilities of machine learning,
  2. have mathmatical tools to formalize machine learning problems, and
  3. build systems by implementing state-of-the-art techniques in machine learning.

Application

Prerequisites

The participants are not expected to have prior exposure to machine learning. However, the program will be most beneficial for participants with some programming experience, and familiarity with basic concepts in probability and statistics, calculus, and linear algebra. Some examples of such concepts include:

Registration

If you are interested in participating, please apply through the link below.
Participation costs for admitted applicants will be fully covered by the NSF RTG grant and no tuition will be charged. Participants are however expected to cover their own transportation and housing costs.

Application for this program is closed.

Schedule

Slides and materials adapted from lectures/slides of David McAllester (TTIC), Greg Shaknarovich (TTIC), David Sontag (MIT), and Nati Srebro (TTIC).

Date Schedule
June 18, Monday 9am-9:30pm
9:30:00-10:55am
11:05-12:30pm
12:30-2:00pm
2:00-5pm

Course setup
Lecture 1.a: Introduction, supervised learning. Instructor: Suriya Gunasekar. [Slides]
Lecture 1.b: Linear regression. Instructor: Karl Stratos. [Slides]
Lunch
Programming
Additional resources: Python numpy tutorial
June 19, Tuesday 9:00am-10:25am
10:35am-noon
noon-1:00pm
1:00pm-3:30pm
3:30-5:00pm
Lecture 2.a: Overfitting, model selection. Instructor: Suriya Gunasekar. [Slides]
Lecture 2.b: Regularization, gradient descent. Instructor: Karl Stratos. [Slides]
Lunch
Programming
Invited Talk - Mathew Walter
June 20, Wednesday 9:00am-10:25am
10:35am-noon
noon-1:00pm
1:00pm-5:00pm
Lecture 3.a: Classification, logistic regression. Instructor: Suriya Gunasekar. [Slides]
Lecture 3.b: Logistic regression continuation, multi-class classification. Instructor: Karl Stratos. [Slides]
Lunch
Programming
June 21, Thursday 9:00am-10:25am
10:35am-noon
noon-1:00pm
1:00pm-3:30pm
3:30-5:00pm
Lecture 4.a: Maximum margin classifier/support vector machines. Instructor: Suriya Gunasekar. [Slides]
Lecture 4.b: Support vector machines continuation. Instructor: Karl Stratos. [Slides]
Lunch
Programming
Invited Talk - Nathan Srebro
June 22, Friday 9:00am-10:25am
10:35am-noon
noon-1:00pm
1:00pm-5:00pm
Lecture 5.a: Generative models, naive Bayes classifier. Instructor: Suriya Gunasekar. [Slides]
Lecture 5.b: Structured classification: hidden Markov models. Instructor: Karl Stratos. [Slides]
Lunch
Programming
June 25, Monday 9:00am-10:25am
10:30am-11:30am
11:30am-12:30pm
12:30pm-2:00pm
2:00pm-5:00pm


Lecture 6.a: Review of week 1, introduction to neural networks. Instructor: Suriya Gunasekar. [Slides]
Invited Talk - Greg Durett (also the TTIC colloquium talk)
Lunch
Lecture 6.b: Backpropagation. Instructor: Karl Stratos. [Slides]
Programming
Additional resources: (a) One of several notes on backpropagation,
(b) Backpropagration demo
June 26, Tuesday 9:00am-10:25am
10:35am-noon
noon-1:00pm
1:00pm-3:30pm
3:30-5:00pm


Lecture 7.a: Optimization and regularization for deep learning. Instructor: Suriya Gunasekar. [Slides]
Lecture 7.b: Special neural network architectures. Instructor: Karl Stratos. [Slides]
Lunch
Programming
Invited Tutorial - Karen Livescu
Additional resource: (a) Notes on (S)GD variants
(b) Demo to play with , (c) CNN lecture notes , (d) RNN notes
June 27, Wednesday 9:00am-10:25am
10:35am-noon
noon-1:00pm
1:00pm-5:00pm
Lecture 8.a: Encoder decoder architectures in NN. Instructor: Karl Stratos.[Slides]
Lecture 8.b: Ensemble methods, boosting. Instructor: Suriya Gunasekar. [Slides]
Lunch
Programming
June 28, Thursday 9:00am-10:25am
10:35am-noon
noon-1:00pm
1:00pm-5:00pm
Lecture 9.a: Unsupervised learning: dimensionality reduction. Instructor: Suriya Gunasekar. [Slides]
Lecture 9.b: Unsupervised learning: EM algorithm. Instructor: Karl Stratos. [Slides]
Lunch
Programming
June 29, Friday 9:00am-10:30am
10:30am-noon
noon-2:00pm
Lecture 10a: Unsupervised leanring: Clustering. Instructor: Karl Stratos. [Slides]
Lecture 10b: Review and conclusion. Instructor: Suriya Gunasekar. [Slides]
Group lunch

Invited Talks

matt nati greg-durett karen
Mathew Walter
TTI Chicago
Nathan Srebro
TTI Chicago
Greg Durett
UT Austin
Karen Livescu
TTI Chicago

People

Instructors

suriya karl mesrob
Suriya Gunasekar
TTI Chicago
Karl Stratos
TTI Chicago
Mesrob I. Ohannessian
TTI Chicago

Teaching Assistants

rebecca pedro kevin
Rebecca Kotsonis Pedro Savarese Kevin Stangl

Advisory Committee

greg mary nati
Gregory Shakhnarovich
TTI Chicago
Mary Silber
CCAM, UChicago
Nathan Srebro
TTI Chicago

Administration

zellencia
Zellencia Harris
CCAM Student Affairs Administrator, UChicago