Regularized last-iterate solvers select the maximum-entropy Nash equilibrium while regret-averaging methods select lower-entropy faces on zero-sum Nash polytopes, verified on analytic testbeds and a 180-game ensemble.
In Proceedings of the 2018 ACM Conference on Economics and Computation (Ithaca, NY, USA) (EC ’18)
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
representative citing papers
Derives first lower bound on γ_t for mean-based algorithms in unknown-horizon bandit settings, proposes two new algorithms, and shows some are also no-regret.
Existence of EF1 and constant-ρ MMS allocations proven for submodular valuations.
For stationary ergodic processes the set of calibration-passing forecast distributions equals the mean-preserving contractions of the conditional distribution, allowing the dynamic game to be solved via static persuasion.
citing papers explorer
-
Which Nash Equilibrium? Solver-Dependent Selection on Zero-Sum Nash Polytopes
Regularized last-iterate solvers select the maximum-entropy Nash equilibrium while regret-averaging methods select lower-entropy faces on zero-sum Nash polytopes, verified on analytic testbeds and a 180-game ensemble.
-
Simultaneous EF1 and approximate MMS allocations for submodular valuations
Existence of EF1 and constant-ρ MMS allocations proven for submodular valuations.