This article deserves way more attention. The Atari DQN paper was a really buzzy thing when it came out. It’s been over four years now and we haven’t seen DQN and similar Deep RL techniques generalize well.
I laughed so hard at the Local Optima example video
This article deserves way more attention. The Atari DQN paper was a really buzzy thing when it came out. It’s been over four years now and we haven’t seen DQN and similar Deep RL techniques generalize well.