FARANE-Q: Fast Parallel and Pipeline Q-Learning Accelerator for Configurable Reinforcement Learning SoC.
Nana SutisnaAndi M. Riyadhus IlmyInfall SyafalniRahmat MulyawanTrio AdionoPublished in: IEEE Access (2023)
Keyphrases
- reinforcement learning
- parallel implementation
- function approximation
- reinforcement learning algorithms
- model free
- learning algorithm
- multi agent
- optimal policy
- action selection
- state space
- temporal difference learning
- relational reinforcement learning
- multi agent reinforcement learning
- reinforcement learning methods
- machine learning
- control problems
- state action space
- stochastic approximation
- optimal control
- learning problems
- eligibility traces
- hardware and software
- dynamic programming
- markov decision processes
- parallel architecture
- learning process
- continuous state and action spaces
- policy iteration
- state action
- computer architecture
- cooperative
- message passing interface
- hierarchical reinforcement learning
- low power
- temporal difference
- parallel computing
- massively parallel
- partially observable markov decision processes
- shared memory
- rl algorithms
- reward function