Neural Network Based Reinforcement Learning Acceleration on FPGA Platforms.
Jiang SuJianxiong LiuDavid B. ThomasPeter Y. K. CheungPublished in: SIGARCH Comput. Archit. News (2016)
Keyphrases
- reinforcement learning
- neural network
- hardware implementation
- high speed
- function approximation
- field programmable gate array
- learning algorithm
- reinforcement learning algorithms
- low cost
- real time
- state space
- multi agent
- software implementation
- supervised learning
- markov decision processes
- data acquisition
- model free
- real time image processing
- action space
- optimal policy
- robotic control
- learning process
- reinforcement learning methods
- pipelined architecture
- power reduction
- low power consumption
- gate array
- single chip
- markov decision process
- computing platform
- temporal difference
- action selection
- signal processing
- machine learning
- dynamic programming