Stochastic First-Order Methods in Data Analysis and Reinforcement Learning
Seminar | September 6 | 4-5 p.m. | 1011 Evans Hall
Mengdi Wang, Princeton University
Stochastic first-order methods provide a basic algorithmic tool for online learning and data analysis. In this talk, we survey several innovative applications including risk-averse optimization, online principal component analysis, dynamic network partition, Markov decision problems and reinforcement learning. We will show that convergence analysis of the stochastic optimization algorithms provide near-optimal sample complexity, run-time complexity and regret analysis in a variety of offline and online learning applications.