0.25
0.5
0.75
1.25
1.5
1.75
2
Escaping Groundhog Day
Published on Jul 28, 20151894 Views
The dominant approaches to reinforcement learning rely on a fixed state-action space and reward function that the agent is trying to maximize. During training, the agent is repeatedly reset to a pred
Related categories
Chapter list
Escaping Groundhog Day00:00
Key Message00:20
Reinforcement Learning01:02
What if something changes?01:48
Groundhog Day Assumptions - 102:35
Groundhog Day Assumptions - 202:56
Groundhog Day Assumptions - 303:18
Still from the movie03:30
Groundhog Day Assumptions - 405:02
Groundhog Day Successes05:11
Between Ground and Figure05:39
Learn what changes06:30
Escaping Groundhog Day07:03
Escaping Assumptions07:21
Related Areas08:05
Problem Generators08:46
Robotics09:00
Minecraft09:43
Reasoning with a Problem Generator10:19
OO-MDPs - 111:03
OO-MDPs - 211:19
OO-MDP Generalization11:33
BURLAP12:01
What we can learn12:42
Learning to Learn13:12
Learning to Plan13:42
Poster14:02
Conclusion - 114:12
Collaborators14:40
Conclusion - 214:46