Learning in an Adversarial World, with Connections to Pricing, Hedging, and Routing

Machine Learning is often viewed through the lens of statistics, where one tries to model or fit a set of data under stochastic conditions. For example, it is typical to assume one's observations were sampled IID. But stochastic assumptions are not always necessary: Blackwell and Hannan in 1950s showed how to construct learning and decision strategies that possess robust guarantees under adversarial conditions. Within this setting the goal of the learner is generally to "minimize regret" against any sequence of inputs. In this talk we lay out the framework, discuss some recent results, and we finish by exploring a few surprising applications and connections: (a) market making in combinatorial prediction markets, (b) routing with limited feedback, and (c) a minimax view of option pricing, with a connection to the classical Black-Scholes valuation model.
Jake received his undergraduate degree in Mathematics from MIT in 2002 and a Master's degree in Computer Science from TTI-C in 2006. He finished a PhD in Computer Science at UC Berkeley, advised by Professor Peter Bartlett, and he is now the Simons Postdoctoral Fellow at University of Pennsylvania with Professor Michael Kearns. Jake's research focuses on the intersection between machine learning, games and markets.