26 citations · 203 across the 41 of their papers we have counts for
8 papers · 1 filter
Difference Rewards Policy Gradients
Jacopo Castellini, Sam Devlin, Frans A. Oliehoek +1
Policy gradient methods have become one of the most popular classes of algorithms for multi-agent reinforcement learning. A key challenge, however, that is not addressed by many of…
The Complexity of Gradient Descent: CLS = PPAD PLS
John Fearnley, Paul W. Goldberg, Alexandros Hollender +1
We study search problems that can be solved by performing Gradient Descent on a bounded convex polytopal domain and show that this class is equal to the intersection of two well-kn…
A faster algorithm for finding Tarski fixed points
John Fearnley, Dömötör Pálvölgyi, Rahul Savani
Dang et al. have given an algorithm that can find a Tarski fixed point in a -dimensional lattice of width using queries. Multiple authors have conjectured th…
A Natural Actor-Critic Algorithm with Downside Risk Constraints
Thomas Spooner, Rahul Savani
Existing work on risk-sensitive reinforcement learning - both for symmetric and downside risk measures - has typically used direct Monte-Carlo estimation of policy gradients. While…
A deep learning approach to identify unhealthy advertisements in street view images
Gregory Palmer, Mark Green, Emma Boyland +3
While outdoor advertisements are common features within towns and cities, they may reinforce social inequalities in health. Vulnerable populations in deprived areas may have greate…
Path Polymatrix Games are PPAD-hard
Argyrios Deligkas, John Fearnley, Rahul Savani
We prove that it is PPAD-hard to compute a Nash equilibrium in a tree polymatrix game with at most twenty actions per player, and path polymatrix game with at most four hundred act…