HAS-RL: A Hierarchical Approximate Scheme Optimized With Reinforcement Learning for NoC-Based NN Accelerators.
Siyue LiShize ZhouYongqi XueWenjie FanTong ChengJinlun JiChenyang DaiWenqing SongQinyu ChenChang GaoLi LiYuxiang FuPublished in: IEEE Trans. Circuits Syst. I Regul. Pap. (2024)
Keyphrases
- reinforcement learning
- neural network
- function approximation
- hierarchical reinforcement learning
- state space
- policy evaluation
- reinforcement learning algorithms
- model free
- multi agent
- markov decision processes
- machine learning
- artificial neural networks
- autonomous learning
- optimal policy
- direct policy search
- nearest neighbor
- dynamic programming
- transfer learning
- temporal difference
- learning process
- knn
- rl algorithms
- reinforcement learning methods
- action selection
- learning algorithm
- temporal difference learning
- action space
- learning agents
- single chip
- control policy
- partially observable markov decision processes
- reward function