Matches in SemOpenAlex for { <https://semopenalex.org/work/W4319996405> ?p ?o ?g. }
Showing items 1 to 71 of
71
with 100 items per page.
- W4319996405 endingPage "2619" @default.
- W4319996405 startingPage "2615" @default.
- W4319996405 abstract "Recently, accelerator/network co-search has shown great promise in reducing the complexity of co-design and achieving higher model accuracy. Unlike the manual design methodology, it automatically learns the optimal network architecture and corresponding accelerator under the given constraints. However, prior works do not consider the direct feedback of searched accelerators on the network search in each step, which leads to low converge speed and sub-optimal solutions. To address this issue, we propose DAN, a reinforcement learning-based framework for fast and accurate accelerator/network co-search. The fundamental idea is to model the co-search as interleaved network-aware accelerator search (AS) and accelerator-aware network search (NS) using separate RL agents, which improves the performance of AS and NS, and encourages a tight interaction between AS and NS. Experimental results show that our proposed method consistently outperforms single-agent based method in terms of converge speed and performance." @default.
- W4319996405 created "2023-02-11" @default.
- W4319996405 creator A5000727470 @default.
- W4319996405 creator A5069255737 @default.
- W4319996405 creator A5074942308 @default.
- W4319996405 date "2023-07-01" @default.
- W4319996405 modified "2023-10-17" @default.
- W4319996405 title "Efficient Accelerator/Network Co-Search With Circular Greedy Reinforcement Learning" @default.
- W4319996405 cites W2162385899 @default.
- W4319996405 cites W2194775991 @default.
- W4319996405 cites W2289252105 @default.
- W4319996405 cites W2606722458 @default.
- W4319996405 cites W2725159389 @default.
- W4319996405 cites W2940862705 @default.
- W4319996405 cites W2964081807 @default.
- W4319996405 cites W2967733054 @default.
- W4319996405 cites W2998732502 @default.
- W4319996405 cites W3016142271 @default.
- W4319996405 cites W3092334294 @default.
- W4319996405 cites W3092581393 @default.
- W4319996405 cites W3096533519 @default.
- W4319996405 cites W3211833531 @default.
- W4319996405 cites W3213229701 @default.
- W4319996405 doi "https://doi.org/10.1109/tcsii.2023.3237912" @default.
- W4319996405 hasPublicationYear "2023" @default.
- W4319996405 type Work @default.
- W4319996405 citedByCount "2" @default.
- W4319996405 countsByYear W43199964052023 @default.
- W4319996405 crossrefType "journal-article" @default.
- W4319996405 hasAuthorship W4319996405A5000727470 @default.
- W4319996405 hasAuthorship W4319996405A5069255737 @default.
- W4319996405 hasAuthorship W4319996405A5074942308 @default.
- W4319996405 hasConcept C11413529 @default.
- W4319996405 hasConcept C154945302 @default.
- W4319996405 hasConcept C173608175 @default.
- W4319996405 hasConcept C193415008 @default.
- W4319996405 hasConcept C31258907 @default.
- W4319996405 hasConcept C41008148 @default.
- W4319996405 hasConcept C51823790 @default.
- W4319996405 hasConcept C68339613 @default.
- W4319996405 hasConcept C97541855 @default.
- W4319996405 hasConceptScore W4319996405C11413529 @default.
- W4319996405 hasConceptScore W4319996405C154945302 @default.
- W4319996405 hasConceptScore W4319996405C173608175 @default.
- W4319996405 hasConceptScore W4319996405C193415008 @default.
- W4319996405 hasConceptScore W4319996405C31258907 @default.
- W4319996405 hasConceptScore W4319996405C41008148 @default.
- W4319996405 hasConceptScore W4319996405C51823790 @default.
- W4319996405 hasConceptScore W4319996405C68339613 @default.
- W4319996405 hasConceptScore W4319996405C97541855 @default.
- W4319996405 hasFunder F4320321133 @default.
- W4319996405 hasIssue "7" @default.
- W4319996405 hasLocation W43199964051 @default.
- W4319996405 hasOpenAccess W4319996405 @default.
- W4319996405 hasPrimaryLocation W43199964051 @default.
- W4319996405 hasRelatedWork W1589376391 @default.
- W4319996405 hasRelatedWork W2051711022 @default.
- W4319996405 hasRelatedWork W2090033344 @default.
- W4319996405 hasRelatedWork W2403150446 @default.
- W4319996405 hasRelatedWork W2959276766 @default.
- W4319996405 hasRelatedWork W3074294383 @default.
- W4319996405 hasRelatedWork W4206669594 @default.
- W4319996405 hasRelatedWork W4327568679 @default.
- W4319996405 hasRelatedWork W4378800968 @default.
- W4319996405 hasRelatedWork W1506942559 @default.
- W4319996405 hasVolume "70" @default.
- W4319996405 isParatext "false" @default.
- W4319996405 isRetracted "false" @default.
- W4319996405 workType "article" @default.