Abstract: This paper studies the online reward poisoning problem, wherein an adversary deliberately manipulates the reward function during training to mislead the learning agent into adopting a ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results