Matches in SemOpenAlex for { <https://semopenalex.org/work/W4282946062> ?p ?o ?g. }
Showing items 1 to 60 of
60
with 100 items per page.
- W4282946062 abstract "Quantization can efficiently reduce the bitwidth of parameters in neural networks and accelerate both inference and data transmission, which is essential for deploying networks with a large parameter scale on resource-limited edge devices. However, most of the existing quantization methods use non-linear function (e.g., tanh, non-linear polynomial function), it is not only difficult to implement on hardware but also occupy plenty of computing resources. In addition, previous quantization methods seldom consider the hardware implementation of operators, which only simulate the quantization accuracy at the algorithm level. To address these problems, we propose novel emph{Trainable Power-of-2 Scale Factors Quantization} (TPSQ) to combine power-of-2 quantization scale factor and trainable clamp interval, to benefit from the advantages of both. Several experiments over current mainstream vision tasks show that the performance of TPSQ surpasses most of the previous quantization methods, these prove the effectiveness of TPSQ. Finally, we implemented an FPGA accelerator for object detection to demonstrate the hardware friendliness of TPSQ. The experimental results show that the peak performance of our system is 300 GOP/s under 300 MHz working frequency, which is 72 times higher than the implementation on an ARM-A9 processor." @default.
- W4282946062 created "2022-06-16" @default.
- W4282946062 creator A5040827219 @default.
- W4282946062 creator A5058767221 @default.
- W4282946062 creator A5072239582 @default.
- W4282946062 creator A5087920560 @default.
- W4282946062 creator A5090067113 @default.
- W4282946062 date "2022-03-18" @default.
- W4282946062 modified "2023-10-16" @default.
- W4282946062 title "Trainable Power-of-2 Scale Factors for Hardware-friendly Network Quantization" @default.
- W4282946062 doi "https://doi.org/10.1109/icccr54399.2022.9790136" @default.
- W4282946062 hasPublicationYear "2022" @default.
- W4282946062 type Work @default.
- W4282946062 citedByCount "0" @default.
- W4282946062 crossrefType "proceedings-article" @default.
- W4282946062 hasAuthorship W4282946062A5040827219 @default.
- W4282946062 hasAuthorship W4282946062A5058767221 @default.
- W4282946062 hasAuthorship W4282946062A5072239582 @default.
- W4282946062 hasAuthorship W4282946062A5087920560 @default.
- W4282946062 hasAuthorship W4282946062A5090067113 @default.
- W4282946062 hasConcept C111919701 @default.
- W4282946062 hasConcept C113775141 @default.
- W4282946062 hasConcept C11413529 @default.
- W4282946062 hasConcept C138236772 @default.
- W4282946062 hasConcept C154945302 @default.
- W4282946062 hasConcept C2776214188 @default.
- W4282946062 hasConcept C28855332 @default.
- W4282946062 hasConcept C41008148 @default.
- W4282946062 hasConcept C42935608 @default.
- W4282946062 hasConcept C50644808 @default.
- W4282946062 hasConcept C79974875 @default.
- W4282946062 hasConcept C9390403 @default.
- W4282946062 hasConceptScore W4282946062C111919701 @default.
- W4282946062 hasConceptScore W4282946062C113775141 @default.
- W4282946062 hasConceptScore W4282946062C11413529 @default.
- W4282946062 hasConceptScore W4282946062C138236772 @default.
- W4282946062 hasConceptScore W4282946062C154945302 @default.
- W4282946062 hasConceptScore W4282946062C2776214188 @default.
- W4282946062 hasConceptScore W4282946062C28855332 @default.
- W4282946062 hasConceptScore W4282946062C41008148 @default.
- W4282946062 hasConceptScore W4282946062C42935608 @default.
- W4282946062 hasConceptScore W4282946062C50644808 @default.
- W4282946062 hasConceptScore W4282946062C79974875 @default.
- W4282946062 hasConceptScore W4282946062C9390403 @default.
- W4282946062 hasLocation W42829460621 @default.
- W4282946062 hasOpenAccess W4282946062 @default.
- W4282946062 hasPrimaryLocation W42829460621 @default.
- W4282946062 hasRelatedWork W2949701228 @default.
- W4282946062 hasRelatedWork W3196944651 @default.
- W4282946062 hasRelatedWork W4220839918 @default.
- W4282946062 hasRelatedWork W4221155876 @default.
- W4282946062 hasRelatedWork W4280607397 @default.
- W4282946062 hasRelatedWork W4295854938 @default.
- W4282946062 hasRelatedWork W4310354755 @default.
- W4282946062 hasRelatedWork W4320023989 @default.
- W4282946062 hasRelatedWork W4323922040 @default.
- W4282946062 hasRelatedWork W4362510878 @default.
- W4282946062 isParatext "false" @default.
- W4282946062 isRetracted "false" @default.
- W4282946062 workType "article" @default.