PDE continuum limits for prediction with expert advice

With Jeff Calder (University of Minnesota)

PDE continuum limits for prediction with expert advice

Prediction with expert advice refers to a class of machine learning problems that is concerned with how to optimally combine advice from multiple experts whose prediction qualities may vary greatly. We study a stock prediction problem with history-dependent experts and an adversarial (worst-case) market, posing the problem as a repeated two-player game. We prove that when the game is played for a long time, the discrete value function converges in the continuum to the solution of a nonlinear parabolic partial differential equation (PDE). This allows us to identify asymptotically optimal strategies for the player and market.

Add to your calendar or Include in your list