mlsanityData to 7 Oct 2026

UWC: Unit-wise Calibration Towards Rapid Network Compression

arXiv 2201.06376 ↗First posted 17 Jan 2022cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

This paper introduces a post-training quantization~(PTQ) method achieving highly efficient Convolutional Neural Network~ (CNN) quantization with high performance. Previous PTQ methods usually reduce compression error via performing layer-by-layer parameters calibration. However, with lower representational ability of extremely compressed parameters (e.g., the bit-width goes less than 4), it is hard to eliminate all the layer-wise errors. This work addresses this issue via proposing a unit-wise feature reconstruction algorithm based on an observation of second order Taylor series expansion of the unit-wise error. It indicates that leveraging the interaction between adjacent layers' parameters could compensate layer-wise errors better. In this paper, we define several adjacent layers as a Basic-Unit, and present a unit-wise post-training algorithm which can minimize quantization error. This method achieves near-original accuracy on ImageNet and COCO when quantizing FP32 models to INT4 and INT3.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 29 of its 34 references that are in the map. First posted on arXiv 17 Jan 2022.

Builds on

29 of 34 references.

References · 29

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  3. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,617
  4. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,451
  5. 2017MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applicationscited by 3,827
  6. 2013Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computationcited by 2,031
  7. 2019Searching for MobileNetV3cited by 1,984
  8. 2016Pruning Filters for Efficient ConvNetscited by 1,129
  9. 2016XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networkscited by 829
  10. 2015Optimizing Neural Networks with Kronecker-factored Approximate Curvaturecited by 776
  11. 2017Channel Pruning for Accelerating Very Deep Neural Networkscited by 697
  12. 2020Up or Down? Adaptive Rounding for Post-Training Quantizationcited by 600
17 more
  1. 2015BinaryConnect: Training Deep Neural Networks with binary weights during propagationscited by 571
  2. 2016DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradientscited by 557
  3. 2021BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstructioncited by 464
  4. 2018PACT: Parameterized Clipping Activation for Quantized Neural Networkscited by 456
  5. 2017ThiNet: A Filter Level Pruning Method for Deep Neural Network Compressioncited by 438
  6. 2018HAQ: Hardware-Aware Automated Quantization with Mixed Precisioncited by 426
  7. 2020ZeroQ: A Novel Zero Shot Quantization Frameworkcited by 265
  8. 2019Differentiable Soft Quantization: Bridging Full-Precision and Low-Bit Neural Networkscited by 234
  9. 2017Data-Driven Sparse Structure Selection for Deep Neural Networkscited by 181
  10. 2019Improving Neural Network Quantization without Retraining using Outlier Channel Splittingcited by 150
  11. 2017Soft Weight-Sharing for Neural Network Compressioncited by 99
  12. 2020Improving Post Training Neural Quantization: Layer-wise Calibration and Integer Programmingcited by 89
  13. 2019Same, Same But Different - Recovering Neural Network Quantization Error Through Weight Factorizationcited by 45
  14. 2020Searching for Low-Bit Weights in Quantized Neural Networkscited by 43
  15. 2018Post-training 4-bit quantization of convolution networks for rapid-deploymentcited by 38
  16. 2020WoodFisher: Efficient Second-Order Approximation for Neural Network Compressioncited by 21
  17. 2016Fixed-point Factorized Networkscited by 5

Built on by

Not cited yet.