FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313142361> ?p ?o ?g. }

• W4313142361 endingPage "13" @default.
• W4313142361 startingPage "1" @default.
• W4313142361 abstract "Long short-term memory (LSTM) is a type of powerful deep neural network that has been widely used in many sequence analysis and modeling applications. However, the large model size problem of LSTM networks make their practical deployment still very challenging, especially for the video recognition tasks that require high-dimensional input data. Aiming to overcome this limitation and fully unlock the potentials of LSTM models, in this paper we propose to perform algorithm and hardware co-design towards high-performance energy-efficient LSTM networks. At algorithm level, we propose to develop <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>fully decomposed hierarchical Tucker (FDHT)</i> structure-based LSTM, namely FDHT-LSTM, which enjoys ultra-low model complexity while still achieving high accuracy. In order to fully reap such attractive algorithmic benefit, we further develop the corresponding customized hardware architecture to support the efficient execution of the proposed FDHT-LSTM model. With the delicate design of memory access scheme, the complicated matrix transformation can be efficiently supported by the underlying hardware without any access conflict in an on-the-fly way. Our evaluation results show that both the proposed ultra-compact FDHT-LSTM models and the corresponding hardware accelerator achieve very high performance. Compared with the state-of-the-art compressed LSTM models, FDHT-LSTM enjoys both order-of-magnitude reduction (more than <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink><tex-math notation=LaTeX>$1000 times$</tex-math></inline-formula> ) in model size and significant accuracy improvement (0.6% to 12.7%) across different video recognition datasets. Meanwhile, compared with the state-of-the-art tensor decomposed model-oriented hardware TIE, our proposed FDHT-LSTM architecture achieve <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink><tex-math notation=LaTeX>$2.5times$</tex-math></inline-formula> , <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink><tex-math notation=LaTeX>$1.46times$</tex-math></inline-formula> and <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink><tex-math notation=LaTeX>$2.41times$</tex-math></inline-formula> increase in throughput, area efficiency and energy efficiency, respectively on LSTM-Youtube workload. For LSTM-UCF workload, our proposed design also outperforms TIE with <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink><tex-math notation=LaTeX>$1.9times$</tex-math></inline-formula> higher throughput, <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink><tex-math notation=LaTeX>$1.83times$</tex-math></inline-formula> higher energy efficiency and comparable area efficiency." @default.
• W4313142361 created "2023-01-06" @default.
• W4313142361 creator A5028255512 @default.
• W4313142361 creator A5054132528 @default.
• W4313142361 creator A5073480376 @default.
• W4313142361 creator A5074398218 @default.
• W4313142361 creator A5083854815 @default.
• W4313142361 date "2022-01-01" @default.
• W4313142361 modified "2023-09-28" @default.
• W4313142361 title "Algorithm and Hardware Co-Design of Energy-Efficient LSTM Networks for Video Recognition with Hierarchical Tucker Tensor Decomposition" @default.
• W4313142361 cites W2005708641 @default.
• W4313142361 cites W2088025933 @default.
• W4313142361 cites W2094756095 @default.
• W4313142361 cites W2100916003 @default.
• W4313142361 cites W2285660444 @default.
• W4313142361 cites W2442974303 @default.
• W4313142361 cites W2585720638 @default.
• W4313142361 cites W2623629680 @default.
• W4313142361 cites W2762381996 @default.
• W4313142361 cites W2791759164 @default.
• W4313142361 cites W2803337603 @default.
• W4313142361 cites W2950025781 @default.
• W4313142361 cites W2951183276 @default.
• W4313142361 cites W2962820060 @default.
• W4313142361 cites W2963838731 @default.
• W4313142361 cites W2966234499 @default.
• W4313142361 cites W2979439447 @default.
• W4313142361 cites W3006942268 @default.
• W4313142361 cites W3176934054 @default.
• W4313142361 cites W4236868170 @default.
• W4313142361 cites W4249932213 @default.
• W4313142361 doi "https://doi.org/10.1109/tc.2022.3212642" @default.
• W4313142361 hasPublicationYear "2022" @default.
• W4313142361 type Work @default.
• W4313142361 citedByCount "1" @default.
• W4313142361 countsByYear W43131423612023 @default.
• W4313142361 crossrefType "journal-article" @default.
• W4313142361 hasAuthorship W4313142361A5028255512 @default.
• W4313142361 hasAuthorship W4313142361A5054132528 @default.
• W4313142361 hasAuthorship W4313142361A5073480376 @default.
• W4313142361 hasAuthorship W4313142361A5074398218 @default.
• W4313142361 hasAuthorship W4313142361A5083854815 @default.
• W4313142361 hasBestOaLocation W43131423612 @default.
• W4313142361 hasConcept C11413529 @default.
• W4313142361 hasConcept C134306372 @default.
• W4313142361 hasConcept C147168706 @default.
• W4313142361 hasConcept C154945302 @default.
• W4313142361 hasConcept C155281189 @default.
• W4313142361 hasConcept C202444582 @default.
• W4313142361 hasConcept C26517878 @default.
• W4313142361 hasConcept C33923547 @default.
• W4313142361 hasConcept C38652104 @default.
• W4313142361 hasConcept C41008148 @default.
• W4313142361 hasConcept C48103436 @default.
• W4313142361 hasConcept C50644808 @default.
• W4313142361 hasConcept C77618280 @default.
• W4313142361 hasConcept C80444323 @default.
• W4313142361 hasConceptScore W4313142361C11413529 @default.
• W4313142361 hasConceptScore W4313142361C134306372 @default.
• W4313142361 hasConceptScore W4313142361C147168706 @default.
• W4313142361 hasConceptScore W4313142361C154945302 @default.
• W4313142361 hasConceptScore W4313142361C155281189 @default.
• W4313142361 hasConceptScore W4313142361C202444582 @default.
• W4313142361 hasConceptScore W4313142361C26517878 @default.
• W4313142361 hasConceptScore W4313142361C33923547 @default.
• W4313142361 hasConceptScore W4313142361C38652104 @default.
• W4313142361 hasConceptScore W4313142361C41008148 @default.
• W4313142361 hasConceptScore W4313142361C48103436 @default.
• W4313142361 hasConceptScore W4313142361C50644808 @default.
• W4313142361 hasConceptScore W4313142361C77618280 @default.
• W4313142361 hasConceptScore W4313142361C80444323 @default.
• W4313142361 hasLocation W43131423611 @default.
• W4313142361 hasLocation W43131423612 @default.
• W4313142361 hasOpenAccess W4313142361 @default.
• W4313142361 hasPrimaryLocation W43131423611 @default.
• W4313142361 hasRelatedWork W1521353230 @default.
• W4313142361 hasRelatedWork W2329452785 @default.
• W4313142361 hasRelatedWork W2356380379 @default.
• W4313142361 hasRelatedWork W2364562957 @default.
• W4313142361 hasRelatedWork W2364808845 @default.
• W4313142361 hasRelatedWork W2373250155 @default.
• W4313142361 hasRelatedWork W2386387936 @default.
• W4313142361 hasRelatedWork W2531206168 @default.
• W4313142361 hasRelatedWork W2800726910 @default.
• W4313142361 hasRelatedWork W2902723393 @default.
• W4313142361 isParatext "false" @default.
• W4313142361 isRetracted "false" @default.
• W4313142361 workType "article" @default.