AMRL: Aggregated Memory For Reinforcement Learning.
Jacob BeckKamil CiosekSam DevlinSebastian TschiatschekCheng ZhangKatja HofmannPublished in: ICLR (2020)
Keyphrases
- reinforcement learning
- function approximation
- memory requirements
- memory usage
- machine learning
- temporal difference
- main memory
- state space
- robotic control
- markov decision processes
- low memory
- learning algorithm
- website
- optimal control
- associative memory
- neural network
- reinforcement learning algorithms
- computing power
- real robot
- markov decision process
- stochastic approximation
- memory size
- multi agent
- optimal policy