Hashing Over Predicted Future Frames for Informed Exploration of Deep Reinforcement Learning.
Haiyan YinSinno Jialin PanPublished in: CoRR (2017)
Keyphrases
- reinforcement learning
- active exploration
- long term
- exploration strategy
- exploration exploitation
- function approximation
- real world
- model based reinforcement learning
- data structure
- dynamic programming
- machine learning
- learning process
- optimal control
- action selection
- temporal difference
- reinforcement learning algorithms
- data sets
- active learning
- video copy detection
- moving objects
- hash table
- nearest neighbor search
- model free
- hamming distance
- hash functions
- state space
- data points
- evolutionary algorithm