1 paper · 1 filter
Kiarash Kazari, Ezzeldin Shereen, György Dán
We address the problem of detecting adversarial attacks against cooperative multi-agent reinforcement learning with continuous action space. We propose a decentralized detector tha…