Pennsylvania State University
Date & Time
November 3, 2023, 11:00 am – 12:00 pm
Title: Rough Path Theory in Statistical Learning
Abstract: In this presentation, I will discuss some applications of rough path theory in statistical learning. First, I will address a finance-related problem, specifically a sequential optimal timing problem related to trading price spreads characterized by mean-reverting behavior. Our innovative approach leverages the power of rough path theory, using the signature of a stochastic process to optimize entry and exit times. Second, I will delve into continuous-time reinforcement learning in rough environments. To effectively model uncertainty and facilitate exploration in these environments, our approach integrates relaxed controls into the decision-making process. We will define and examine the fundamental properties of the pathwise version of this rough control problem, shedding light on its potential to enhance reinforcement learning strategies.