SFV: reinforcement learning of physical skills from videos.
Xue Bin PengAngjoo KanazawaJitendra MalikPieter AbbeelSergey LevinePublished in: ACM Trans. Graph. (2018)
Keyphrases
- reinforcement learning
- function approximation
- video sequences
- markov decision processes
- machine learning
- high level knowledge
- temporal difference
- video analysis
- video content
- real world
- video frames
- human activities
- key frames
- video data
- supervised learning
- optimal policy
- moving camera
- lifelong learning
- optimal control
- policy search
- video search
- reinforcement learning algorithms
- learning process
- state space
- dynamic scenes
- activity recognition