E2HRL: An Energy-efficient Hardware Accelerator for Hierarchical Deep Reinforcement Learning.
Aidin ShiriUtteja KallakuriHasib-Al RashidBharat PrakashNicholas R. WaytowichTim OatesTinoosh MohseninPublished in: ACM Trans. Design Autom. Electr. Syst. (2022)
Keyphrases
- reinforcement learning
- field programmable gate array
- hierarchical reinforcement learning
- low cost
- hardware and software
- computer systems
- real time
- multi agent
- hardware implementation
- sensor networks
- multiple layers
- unsupervised learning
- computing power
- embedded systems
- coarse to fine
- optimal policy
- dynamic programming
- wireless sensor networks
- state space
- parallel implementation
- model free
- learning process
- hierarchical model
- transfer learning
- policy search
- deep belief networks
- energy efficient
- hierarchical structure
- peer to peer
- active learning
- machine learning