Matches in SemOpenAlex for { <https://semopenalex.org/work/W3089703283> ?p ?o ?g. }
- W3089703283 abstract "The use of deep learning has grown at an exponential rate, giving rise to numerous specialized hardware and software systems for deep learning. Because the design space of deep learning software stacks and hardware accelerators is diverse and vast, prior work considers software optimizations separately from hardware architectures, effectively reducing the search space. Unfortunately, this bifurcated approach means that many profitable design points are never explored. This paper instead casts the problem as hardware/software co-design, with the goal of automatically identifying desirable points in the joint design space. The key to our solution is a new constrained Bayesian optimization framework that avoids invalid solutions by exploiting the highly constrained features of this design space, which are semi-continuous/semi-discrete. We evaluate our optimization framework by applying it to a variety of neural models, improving the energy-delay product by 18% (ResNet) and 40% (DQN) over hand-tuned state-of-the-art systems, as well as demonstrating strong results on other neural network architectures, such as MLPs and Transformers." @default.
- W3089703283 created "2020-10-08" @default.
- W3089703283 creator A5026208886 @default.
- W3089703283 creator A5026546407 @default.
- W3089703283 creator A5052763112 @default.
- W3089703283 creator A5081588692 @default.
- W3089703283 creator A5086314063 @default.
- W3089703283 date "2020-10-05" @default.
- W3089703283 modified "2023-09-26" @default.
- W3089703283 title "Learned Hardware/Software Co-Design of Neural Accelerators" @default.
- W3089703283 cites W1509443229 @default.
- W3089703283 cites W1510052597 @default.
- W3089703283 cites W1522301498 @default.
- W3089703283 cites W1693986406 @default.
- W3089703283 cites W1746819321 @default.
- W3089703283 cites W1757796397 @default.
- W3089703283 cites W2034761517 @default.
- W3089703283 cites W2039522160 @default.
- W3089703283 cites W2048266589 @default.
- W3089703283 cites W2067523571 @default.
- W3089703283 cites W2099201756 @default.
- W3089703283 cites W2111066976 @default.
- W3089703283 cites W2125203716 @default.
- W3089703283 cites W2131241448 @default.
- W3089703283 cites W2135653967 @default.
- W3089703283 cites W2152839228 @default.
- W3089703283 cites W2167789032 @default.
- W3089703283 cites W2192203593 @default.
- W3089703283 cites W2194775991 @default.
- W3089703283 cites W2442974303 @default.
- W3089703283 cites W2471164860 @default.
- W3089703283 cites W2518511512 @default.
- W3089703283 cites W2565851976 @default.
- W3089703283 cites W2605347906 @default.
- W3089703283 cites W2606722458 @default.
- W3089703283 cites W2625457103 @default.
- W3089703283 cites W2912012512 @default.
- W3089703283 cites W2940862705 @default.
- W3089703283 cites W2945146780 @default.
- W3089703283 cites W2953212265 @default.
- W3089703283 cites W2955425717 @default.
- W3089703283 cites W2962687950 @default.
- W3089703283 cites W2963403868 @default.
- W3089703283 cites W2963805801 @default.
- W3089703283 cites W2963960923 @default.
- W3089703283 cites W2976305354 @default.
- W3089703283 cites W2980104813 @default.
- W3089703283 cites W3021954012 @default.
- W3089703283 cites W3102385154 @default.
- W3089703283 cites W3124229194 @default.
- W3089703283 cites W60686164 @default.
- W3089703283 doi "https://doi.org/10.48550/arxiv.2010.02075" @default.
- W3089703283 hasPublicationYear "2020" @default.
- W3089703283 type Work @default.
- W3089703283 sameAs 3089703283 @default.
- W3089703283 citedByCount "4" @default.
- W3089703283 countsByYear W30897032832021 @default.
- W3089703283 crossrefType "posted-content" @default.
- W3089703283 hasAuthorship W3089703283A5026208886 @default.
- W3089703283 hasAuthorship W3089703283A5026546407 @default.
- W3089703283 hasAuthorship W3089703283A5052763112 @default.
- W3089703283 hasAuthorship W3089703283A5081588692 @default.
- W3089703283 hasAuthorship W3089703283A5086314063 @default.
- W3089703283 hasBestOaLocation W30897032831 @default.
- W3089703283 hasConcept C108583219 @default.
- W3089703283 hasConcept C113775141 @default.
- W3089703283 hasConcept C118524514 @default.
- W3089703283 hasConcept C13164978 @default.
- W3089703283 hasConcept C149635348 @default.
- W3089703283 hasConcept C154945302 @default.
- W3089703283 hasConcept C199360897 @default.
- W3089703283 hasConcept C2776221188 @default.
- W3089703283 hasConcept C2777904410 @default.
- W3089703283 hasConcept C2778049539 @default.
- W3089703283 hasConcept C41008148 @default.
- W3089703283 hasConcept C50644808 @default.
- W3089703283 hasConcept C65232700 @default.
- W3089703283 hasConcept C9390403 @default.
- W3089703283 hasConceptScore W3089703283C108583219 @default.
- W3089703283 hasConceptScore W3089703283C113775141 @default.
- W3089703283 hasConceptScore W3089703283C118524514 @default.
- W3089703283 hasConceptScore W3089703283C13164978 @default.
- W3089703283 hasConceptScore W3089703283C149635348 @default.
- W3089703283 hasConceptScore W3089703283C154945302 @default.
- W3089703283 hasConceptScore W3089703283C199360897 @default.
- W3089703283 hasConceptScore W3089703283C2776221188 @default.
- W3089703283 hasConceptScore W3089703283C2777904410 @default.
- W3089703283 hasConceptScore W3089703283C2778049539 @default.
- W3089703283 hasConceptScore W3089703283C41008148 @default.
- W3089703283 hasConceptScore W3089703283C50644808 @default.
- W3089703283 hasConceptScore W3089703283C65232700 @default.
- W3089703283 hasConceptScore W3089703283C9390403 @default.
- W3089703283 hasLocation W30897032831 @default.
- W3089703283 hasOpenAccess W3089703283 @default.
- W3089703283 hasPrimaryLocation W30897032831 @default.
- W3089703283 hasRelatedWork W2098735319 @default.
- W3089703283 hasRelatedWork W2103962833 @default.
- W3089703283 hasRelatedWork W2110674252 @default.
- W3089703283 hasRelatedWork W2615328932 @default.
- W3089703283 hasRelatedWork W2780340867 @default.