Action-Gap Phenomenon in Reinforcement Learning thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Action-Gap Phenomenon in Reinforcement Learning

Published on Sep 06, 20122994 Views

Many practitioners of reinforcement learning problems have observed that oftentimes the performance of the agent reaches very close to the optimal performance even though the estimated (action-)value

Related categories

Chapter list

Action-Gap Phenomenon in Reinforcement Learning00:00
Easy choice!00:12
Not a big deal if we choose the wrong one!00:46
Finite-action discounted MDP with general state space.01:27