FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3170004439> ?p ?o ?g. }

• W3170004439 endingPage "25" @default.
• W3170004439 startingPage "1" @default.
• W3170004439 abstract "Model compression through quantization is commonly applied to convolutional neural networks (CNNs) deployed on compute and memory-constrained embedded platforms. Different layers of the CNN can have varying degrees of numerical precision for both weights and activations, resulting in a large search space. Together with the hardware (HW) design space, the challenge of finding the globally optimal HW-CNN combination for a given application becomes daunting. To this end, we propose HW-FlowQ, a systematic approach that enables the co-design of the target hardware platform and the compressed CNN model through quantization. The search space is viewed at three levels of abstraction, allowing for an iterative approach for narrowing down the solution space before reaching a high-fidelity CNN hardware modeling tool, capable of capturing the effects of mixed-precision quantization strategies on different hardware architectures (processing unit counts, memory levels, cost models, dataflows) and two types of computation engines (bit-parallel vectorized, bit-serial). To combine both worlds, a multi-objective non-dominated sorting genetic algorithm (NSGA-II) is leveraged to establish a Pareto-optimal set of quantization strategies for the target HW-metrics at each abstraction level. HW-FlowQ detects optima in a discrete search space and maximizes the task-related accuracy of the underlying CNN while minimizing hardware-related costs. The Pareto-front approach keeps the design space open to a range of non-dominated solutions before refining the design to a more detailed level of abstraction. With equivalent prediction accuracy, we improve the energy and latency by 20% and 45% respectively for ResNet56 compared to existing mixed-precision search methods." @default.
• W3170004439 created "2021-06-22" @default.
• W3170004439 creator A5005732789 @default.
• W3170004439 creator A5026016005 @default.
• W3170004439 creator A5034106438 @default.
• W3170004439 creator A5035028804 @default.
• W3170004439 creator A5047063939 @default.
• W3170004439 creator A5056231964 @default.
• W3170004439 creator A5057027966 @default.
• W3170004439 creator A5060720819 @default.
• W3170004439 creator A5083288487 @default.
• W3170004439 creator A5085220085 @default.
• W3170004439 date "2021-09-17" @default.
• W3170004439 modified "2023-09-26" @default.
• W3170004439 title "HW-FlowQ: A Multi-Abstraction Level HW-CNN Co-design Quantization Methodology" @default.
• W3170004439 cites W1999085092 @default.
• W3170004439 cites W2117539524 @default.
• W3170004439 cites W2126105956 @default.
• W3170004439 cites W2194775991 @default.
• W3170004439 cites W2300242332 @default.
• W3170004439 cites W2340897893 @default.
• W3170004439 cites W2504676333 @default.
• W3170004439 cites W2605347906 @default.
• W3170004439 cites W2618530766 @default.
• W3170004439 cites W2719597717 @default.
• W3170004439 cites W2795915628 @default.
• W3170004439 cites W2886851211 @default.
• W3170004439 cites W2891946740 @default.
• W3170004439 cites W2899645952 @default.
• W3170004439 cites W2906043559 @default.
• W3170004439 cites W2935331687 @default.
• W3170004439 cites W2940862705 @default.
• W3170004439 cites W2952110101 @default.
• W3170004439 cites W2963140066 @default.
• W3170004439 cites W2980137827 @default.
• W3170004439 cites W2982041622 @default.
• W3170004439 cites W2982479999 @default.
• W3170004439 cites W3001665736 @default.
• W3170004439 cites W3016142271 @default.
• W3170004439 cites W3035332806 @default.
• W3170004439 cites W4288083528 @default.
• W3170004439 doi "https://doi.org/10.1145/3476997" @default.
• W3170004439 hasPublicationYear "2021" @default.
• W3170004439 type Work @default.
• W3170004439 sameAs 3170004439 @default.
• W3170004439 citedByCount "4" @default.
• W3170004439 countsByYear W31700044392022 @default.
• W3170004439 countsByYear W31700044392023 @default.
• W3170004439 crossrefType "journal-article" @default.
• W3170004439 hasAuthorship W3170004439A5005732789 @default.
• W3170004439 hasAuthorship W3170004439A5026016005 @default.
• W3170004439 hasAuthorship W3170004439A5034106438 @default.
• W3170004439 hasAuthorship W3170004439A5035028804 @default.
• W3170004439 hasAuthorship W3170004439A5047063939 @default.
• W3170004439 hasAuthorship W3170004439A5056231964 @default.
• W3170004439 hasAuthorship W3170004439A5057027966 @default.
• W3170004439 hasAuthorship W3170004439A5060720819 @default.
• W3170004439 hasAuthorship W3170004439A5083288487 @default.
• W3170004439 hasAuthorship W3170004439A5085220085 @default.
• W3170004439 hasConcept C111472728 @default.
• W3170004439 hasConcept C113775141 @default.
• W3170004439 hasConcept C11413529 @default.
• W3170004439 hasConcept C118524514 @default.
• W3170004439 hasConcept C124304363 @default.
• W3170004439 hasConcept C138885662 @default.
• W3170004439 hasConcept C149635348 @default.
• W3170004439 hasConcept C154945302 @default.
• W3170004439 hasConcept C173608175 @default.
• W3170004439 hasConcept C2776221188 @default.
• W3170004439 hasConcept C28855332 @default.
• W3170004439 hasConcept C41008148 @default.
• W3170004439 hasConcept C81363708 @default.
• W3170004439 hasConcept C9390403 @default.
• W3170004439 hasConceptScore W3170004439C111472728 @default.
• W3170004439 hasConceptScore W3170004439C113775141 @default.
• W3170004439 hasConceptScore W3170004439C11413529 @default.
• W3170004439 hasConceptScore W3170004439C118524514 @default.
• W3170004439 hasConceptScore W3170004439C124304363 @default.
• W3170004439 hasConceptScore W3170004439C138885662 @default.
• W3170004439 hasConceptScore W3170004439C149635348 @default.
• W3170004439 hasConceptScore W3170004439C154945302 @default.
• W3170004439 hasConceptScore W3170004439C173608175 @default.
• W3170004439 hasConceptScore W3170004439C2776221188 @default.
• W3170004439 hasConceptScore W3170004439C28855332 @default.
• W3170004439 hasConceptScore W3170004439C41008148 @default.
• W3170004439 hasConceptScore W3170004439C81363708 @default.
• W3170004439 hasConceptScore W3170004439C9390403 @default.
• W3170004439 hasIssue "5s" @default.
• W3170004439 hasLocation W31700044391 @default.
• W3170004439 hasOpenAccess W3170004439 @default.
• W3170004439 hasPrimaryLocation W31700044391 @default.
• W3170004439 hasRelatedWork W1471324948 @default.
• W3170004439 hasRelatedWork W1508156141 @default.
• W3170004439 hasRelatedWork W1997145140 @default.
• W3170004439 hasRelatedWork W2034384303 @default.
• W3170004439 hasRelatedWork W2099940607 @default.
• W3170004439 hasRelatedWork W2108386578 @default.
• W3170004439 hasRelatedWork W2187267005 @default.

• next