Matches in SemOpenAlex for { <https://semopenalex.org/work/W2914725495> ?p ?o ?g. }
Showing items 1 to 79 of
79
with 100 items per page.
- W2914725495 abstract "State-of-the-art hardware accelerators for large scale CNNs face two challenges: high computation complexity of convolution, and high on-chip memory consumption by weight kernels. Two techniques have been proposed in the literature to address these challenges: frequency domain convolution and space domain fixed-point quantization. In this paper, we propose frequency domain quantization schemes to achieve high throughput CNN inference on FPGAs. We first analyze the impact of quantization bit width on the accuracy of a frequency domain CNN, via the metric of Signal-to-Quantization-Noise-Ratio (SQNR). Taking advantage of the reconfigurability of FPGAs, we design a statically-reconfigurable and a dynamically-reconfigurable architecture for the quantized convolutional layers. Then, based on the SQNR analysis, we propose quantization schemes for both types of architectures, achieving optimal tradeoff between throughput and accuracy. The proposed quantizer allocates the number of bits for each convolutional layer under various design constraints, including overall SQNR, available DSP resources, on-chip memory and off-chip bandwidth. Experiments on AlexNet show that our designs improve the CNN inference throughput by 1.45$times$ to 8.44$times$, with negligible (< 0.5%) loss in accuracy." @default.
- W2914725495 created "2019-02-21" @default.
- W2914725495 creator A5014103790 @default.
- W2914725495 creator A5033166029 @default.
- W2914725495 creator A5045357286 @default.
- W2914725495 creator A5084855872 @default.
- W2914725495 date "2018-12-01" @default.
- W2914725495 modified "2023-10-10" @default.
- W2914725495 title "Throughput-Optimized Frequency Domain CNN with Fixed-Point Quantization on FPGA" @default.
- W2914725495 cites W1969057818 @default.
- W2914725495 cites W2163491234 @default.
- W2914725495 cites W2171656599 @default.
- W2914725495 cites W2233116163 @default.
- W2914725495 cites W2584616277 @default.
- W2914725495 cites W2786216825 @default.
- W2914725495 cites W2786845740 @default.
- W2914725495 cites W2792503273 @default.
- W2914725495 doi "https://doi.org/10.1109/reconfig.2018.8641716" @default.
- W2914725495 hasPublicationYear "2018" @default.
- W2914725495 type Work @default.
- W2914725495 sameAs 2914725495 @default.
- W2914725495 citedByCount "6" @default.
- W2914725495 countsByYear W29147254952019 @default.
- W2914725495 countsByYear W29147254952022 @default.
- W2914725495 countsByYear W29147254952023 @default.
- W2914725495 crossrefType "proceedings-article" @default.
- W2914725495 hasAuthorship W2914725495A5014103790 @default.
- W2914725495 hasAuthorship W2914725495A5033166029 @default.
- W2914725495 hasAuthorship W2914725495A5045357286 @default.
- W2914725495 hasAuthorship W2914725495A5084855872 @default.
- W2914725495 hasConcept C11413529 @default.
- W2914725495 hasConcept C134306372 @default.
- W2914725495 hasConcept C157764524 @default.
- W2914725495 hasConcept C163973906 @default.
- W2914725495 hasConcept C173608175 @default.
- W2914725495 hasConcept C19118579 @default.
- W2914725495 hasConcept C28855332 @default.
- W2914725495 hasConcept C31972630 @default.
- W2914725495 hasConcept C33923547 @default.
- W2914725495 hasConcept C41008148 @default.
- W2914725495 hasConcept C42935608 @default.
- W2914725495 hasConcept C555944384 @default.
- W2914725495 hasConcept C61445026 @default.
- W2914725495 hasConcept C76155785 @default.
- W2914725495 hasConcept C84211073 @default.
- W2914725495 hasConcept C9390403 @default.
- W2914725495 hasConceptScore W2914725495C11413529 @default.
- W2914725495 hasConceptScore W2914725495C134306372 @default.
- W2914725495 hasConceptScore W2914725495C157764524 @default.
- W2914725495 hasConceptScore W2914725495C163973906 @default.
- W2914725495 hasConceptScore W2914725495C173608175 @default.
- W2914725495 hasConceptScore W2914725495C19118579 @default.
- W2914725495 hasConceptScore W2914725495C28855332 @default.
- W2914725495 hasConceptScore W2914725495C31972630 @default.
- W2914725495 hasConceptScore W2914725495C33923547 @default.
- W2914725495 hasConceptScore W2914725495C41008148 @default.
- W2914725495 hasConceptScore W2914725495C42935608 @default.
- W2914725495 hasConceptScore W2914725495C555944384 @default.
- W2914725495 hasConceptScore W2914725495C61445026 @default.
- W2914725495 hasConceptScore W2914725495C76155785 @default.
- W2914725495 hasConceptScore W2914725495C84211073 @default.
- W2914725495 hasConceptScore W2914725495C9390403 @default.
- W2914725495 hasLocation W29147254951 @default.
- W2914725495 hasOpenAccess W2914725495 @default.
- W2914725495 hasPrimaryLocation W29147254951 @default.
- W2914725495 hasRelatedWork W1482661175 @default.
- W2914725495 hasRelatedWork W1507077400 @default.
- W2914725495 hasRelatedWork W1514208968 @default.
- W2914725495 hasRelatedWork W1793682442 @default.
- W2914725495 hasRelatedWork W1983878430 @default.
- W2914725495 hasRelatedWork W2022696940 @default.
- W2914725495 hasRelatedWork W2096545584 @default.
- W2914725495 hasRelatedWork W2320205417 @default.
- W2914725495 hasRelatedWork W3119699820 @default.
- W2914725495 hasRelatedWork W4221148635 @default.
- W2914725495 isParatext "false" @default.
- W2914725495 isRetracted "false" @default.
- W2914725495 magId "2914725495" @default.
- W2914725495 workType "article" @default.