CSC 581A/482A Project Details

Who is presenting what

Paper	Presenter
Online Mirror Descent and Dual Averaging: Keeping Pace in the Dynamic Case	Quan Nguyen
Adaptive Subgradient Methods for Online Learning and Stochastic Optimization	Ezra MacDonald
Logarithmic Regret Algorithms for Online Convex Optimization	Yifeng Liu
Efficient Algorithms for Online Decision Problems	Ali Mortazavi
Online Sparse Reinforcement Learning	Homayoun Honari
Anytime Online-to-Batch, Optimism and Acceleration	Mica Grant-Hagen
Online Bandit Learning Against an Adaptive Adversary: From Regret to Policy Regret	Andrea Nguyen

About the course project

As part of this course, every student will complete a solo project with a theoretical focus. The idea of the project is to select a paper or two and present the main ideas of the paper(s) to the class. There also will be a brief (3-5 page report) due on Friday, April 14th at 11:59pm. Please submit via Brightspace, where you can also find a brief description of what to submit.

To help get you started thinking about papers, here is a list of papers in online learning:

Abernethy, Hazan, Rakhlin. Competing in the dark: An efficient algorithm for bandit linear optimization. COLT, 2008.
Alon, Ben-Eliezer, Dagan, Moran, Naor, and Yogev. Adversarial laws of large numbers and optimal regret in online classification. STOC, 2021.
Duchi, Hazan, and Singer. Adaptive subgradient methods for online learning and stochastic optimization. JMLR, 2011.
Blum, and Mansour. From external to internal regret. JMLR, 2007.
Hart and Mas-Colell. A simple adaptive procedure leading to correlated equilibrium. Econometrica, 2000.
Hazan and Seshadhri. Adaptive algorithms for online optimization. ICML, 2019 (full version).
Arora, Dekel, and Tewari. Online bandit learning against an adaptive adversary: from regret to policy regret. ICML, 2012.
Anava, Hazan, and Mannor. Online learning for adversaries with memory: price of past mistakes. NIPS, 2015.
Foster, Kale, Luo, Mohri, and Sridharan. Logistic regression: The importance of being improper. COLT, 2018.
Agarwal, Kale, and Zimmert. Efficient methods for online multiclass logistic regression. ALT, 2022.
Luo, Wei, and Zheng. Efficient online portfolio with logarithmic regret. NeurIPS, 2018.
Jézéquel, Ostrovskii, and Gaillard. Efficient and near-optimal online portfolio selection. 2022 (arXiv).
Cutkosky. Anytime online-to-batch, optimism and acceleration. ICML, 2019.
Hazan, Agarwal, Kale. Logarithmic regret algorithms for online convex optimization. Machine Learning, 2007.

Presentation

For each project, there will be a 20 to 25 minute presentation (depending on the time available) to discuss your findings, plus a few minutes for questions. These presentations will happen in the last week of lectures.

Submission

Each of you also will submit a 3-5 page report, not including references. The maximum length is flexible depending on your project. For your report, please use the LaTeX style file and template below (this is just to get the geometry right, e.g. margins, font size, line spacing):