Skip to yearly menu bar Skip to main content


Spotlight Poster

Generative Adversarial Inverse Multiagent Learning

Denizalp Goktas · Amy Greenwald · Sadie Zhao · Alec Koppel · Sumitra Ganesh

Halle B
[ ]
Tue 7 May 1:45 a.m. PDT — 3:45 a.m. PDT
 
Spotlight presentation:

Abstract:

In this paper, we study inverse game theory (resp. inverse multiagent learning) inwhich the goal is to find parameters of a game’s payoff functions for which theexpected (resp. sampled) behavior is an equilibrium. We formulate these problemsas a generative-adversarial (i.e., min-max) optimization problem, based on whichwe develop polynomial-time algorithms the solve them, the former of whichrelies on an exact first-order oracle, and the latter, a stochastic one. We extendour approach to solve inverse multiagent apprenticeship learning in polynomialtime and number of samples, where we seek a simulacrum, i.e., parameters andan associated equilibrium, which replicate observations in expectation. We findthat our approach outperforms other widely-used methods in predicting prices inSpanish electricity markets based on time-series data.

Chat is not available.