Machine Learning Reading Group

The Machine Learning Reading Group is a biweekly reading group for discussing theory-oriented machine learning research papers and topics. Since it's a reading group and not a simply a talk, all attendees should read a session's material in advance so that we can critically discuss core ideas.

Meetings from previous years (back to current page)

Spring 2020
date lead topic
May 6 Nishant The Algorithmic Foundations of Differential Privacy, Pages 37–52 (Sections 3.4–3.5.2)
April 27 Bingshan The Algorithmic Foundations of Differential Privacy, Pages 11–34
February 19 Hamid Reconciling modern machine learning practice and the bias-variance trade-off
January 23 Sajjad Envy-free classification

Fall 2019
date lead topic
November 20 Sharoff A reduction of imitation learning and structured prediction to no-regret online learning
September 4 Sajjad On preserving non-discrimination when combining expert advice

Summer 2019
date lead topic
August 22 Bingshan Building bridges: Viewing active learning from the multi-armed bandit lens
August 8 Hamid Active Learning: Part II - Active learning with disagreement graphs
August 1 Hamid Active Learning: Part I - Importance weighted active learning
July 11 Sharoff Sequential transfer in multi-armed bandit with finite set of models
June 20 Sajjad Sample compression schemes for VC classes
June 6 Hamid Learnability can be undecidable

Spring 2019
date lead topic
May 10 Mahdi Influence maximization with bandits
Apr 12 Bingshan Near-optimal regret bounds for Thompson sampling
Mar 15 Sharoff Exploiting easy data in online optimization
Feb 15 Sajjad Regret to the best vs. regret to the average
Feb 1 Hamid Online convex programming and generalized infinitesimal gradient ascent

Fall 2018
date lead topic
Nov 23 Hamid Regret bounds for lifelong learning
Nov 16 Sharoff Regret bound for the stochastic multi-armed bandit problem
Nov 9 Sajjad Agnostic online learnability
Nov 5 Bingshan Matroid bandits: Fast combinatorial optimization with learning