Learning Correlated Equilibrium Via Neural Network Regret
The results show that this bias is rapidly eroded under subsequent adversarial training on heterogeneous games, where play reverts to mixed or risk-sensitive equilibria.
Khushi Sampat
· 0 citations