Faculty Recruiting Make a Gift

From Optimization to Equilibration: Understanding an Emerging Paradigm in AI and ML

12 Dec
Wednesday, 12/12/2018 10:00am to 12:00pm
CS Building, Room 150/151
Ph.D. Thesis Defense
Speaker: Ian Gemp

From Optimization to Equilibration: Understanding an Emerging Paradigm in AI and ML

Abstract: Artificial Intelligence (AI) focuses on the design of agents that act rationally. The Maximum Expected Utility (MEU) principle formalizes the idea of a rational agent as an optimization problem. This principle has pulled optimization to the center of attention in AI and Machine Learning (ML), however, a new paradigm is emerging. Many existing algorithms such as those in Reinforcement Learning (RL) or inference in graphical models can be viewed as solving equilibrium problems. Moreover, some recent ML models, specifically generative adversarial networks (GANs) and its variants, are now explicitly formulated as equilibrium problems.
In this thesis, we aim to advance our understanding of equilibrium problems so as to improve state-of-the-art in these and related domains. In particular, we focus on building a framework for solving equilibrium problems \emph{online}, training a GAN to fit a normal distribution to data, improving state-of-the-art image generation with heuristics derived from theory, and developing tools for analyzing the complex dynamics of training.

Advisor: Sridhar Mahadevan