Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Ni, Yang | - |
dc.contributor.author | Abraham, Danny | - |
dc.contributor.author | Issa, Mariam | - |
dc.contributor.author | Kim, Yeseong | - |
dc.contributor.author | Mercati, Pietro | - |
dc.contributor.author | Imani, Mohsen | - |
dc.date.accessioned | 2024-02-08T22:40:11Z | - |
dc.date.available | 2024-02-08T22:40:11Z | - |
dc.date.created | 2023-07-14 | - |
dc.date.issued | 2023-06-06 | - |
dc.identifier.isbn | 9798400701252 | - |
dc.identifier.uri | http://hdl.handle.net/20.500.11750/47906 | - |
dc.description.abstract | Reinforcement Learning (RL) has opened up new opportunities to enhance existing smart systems that generally include a complex decision-making process. However, modern RL algorithms, e.g., Deep Q-Networks (DQN), are based on deep neural networks, resulting in high computational costs. In this paper, we propose QHD, an off-policy value-based Hyperdimensional Reinforcement Learning, that mimics brain properties toward robust and real-time learning. QHD relies on a lightweight brain-inspired model to learn an optimal policy in an unknown environment. On both desktop and power-limited embedded platforms, QHD achieves significantly better overall efficiency than DQN while providing higher or comparable rewards. QHD is also suitable for highly-efficient reinforcement learning with great potential for online and real-time learning. Our solution supports a small experience replay batch size that provides 12.3 times speedup compared to DQN while ensuring minimal quality loss. Our evaluation shows QHD capability for real-time learning, providing 34.6 times speedup and significantly better quality of learning than DQN. © 2023 Owner/Author. | - |
dc.language | English | - |
dc.publisher | ACM Special Interest Group on Design Automation (SIGDA), IEEE Council on Electronic Design Automation (CEDA) | - |
dc.title | Efficient Off-Policy Reinforcement Learning via Brain-Inspired Computing | - |
dc.type | Conference Paper | - |
dc.identifier.doi | 10.1145/3583781.3590298 | - |
dc.identifier.scopusid | 2-s2.0-85163196609 | - |
dc.identifier.bibliographicCitation | ACM Great Lakes Symposium on VLSI, GLSVLSI 2023, pp.449 - 453 | - |
dc.citation.conferencePlace | US | - |
dc.citation.conferencePlace | Knoxville | - |
dc.citation.endPage | 453 | - |
dc.citation.startPage | 449 | - |
dc.citation.title | ACM Great Lakes Symposium on VLSI, GLSVLSI 2023 | - |
There are no files associated with this item.