Adversarial Training Should Be Cast as a Non-Zero-Sum Game.
Alexander RobeyFabian LatorreGeorge J. PappasHamed HassaniVolkan CevherPublished in: ICLR (2024)
Keyphrases
- game theory
- stochastic games
- nash equilibria
- nash equilibrium
- boolean games
- perfect information
- serious games
- games with incomplete information
- multi agent
- repeated games
- game theoretic
- supervised learning
- training set
- data sets
- multiagent reinforcement learning
- cooperative
- multi agent systems
- game development
- online learning
- incomplete information
- training phase
- training algorithm
- video games
- monte carlo tree search
- training examples
- monte carlo
- test set