1 paper · 1 filter
Trevor A. McInroe, Michael Spurrier, Jennifer Sieber +1
We analyze the hidden activations of neural network policies of deep reinforcement learning (RL) agents and show, empirically, that it's possible to know a priori if a state repres…