Training Quantized Neural Networks With a Full-Precision Auxiliary Module
Open accessThe DOI matches its Crossref record; 23 of the 23 references it cites were found in a registry or an index.
Abstract
In this paper, we seek to tackle a challenge in training low-precision networks: the notorious difficulty in propagating gradient through a low-precision network due to the non-differentiable quantization function. We propose a solution by training the low-precision network with a full-precision auxiliary module. Specifically, during training, we construct a mix-precision network by augmenting the original low-precision network with the full precision auxiliary module. Then the augmented mix-precision network and the low-precision network are jointly optimized. This strategy creates additional full-precision routes to update the parameters of the low-precision model, thus making the gradient back-propagates more easily. At the inference time, we discard the auxiliary module without introducing any computational complexity to the low-precision network. We evaluate the proposed method on image classification and object detection over various quantization approaches and show consistent performance increase. In particular, we achieve near lossless performance to the full-precision model by using a 4-bit detector, which is of great practical value.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946
Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2577031 · Cited by 73,472
You Only Look Once: Unified, Real-Time Object Detection
Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.91 · Cited by 53,803
Feature Pyramid Networks for Object Detection
Lin, T.-Y., Dollar, P., Girshick, R., He, K., Hariharan, B., & Belongie, S. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.106 · Cited by 30,939
XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks
Rastegari, M., Ordonez, V., Redmon, J., & Farhadi, A. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46493-0_32 · Cited by 4,408
Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference
Jacob, B., Kligys, S., Chen, B., Zhu, M., Tang, M., Howard, A., Adam, H., & Kalenichenko, D. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00286 · Cited by 4,114
MnasNet: Platform-Aware Neural Architecture Search for Mobile
Tan, M., Chen, B., Pang, R., Vasudevan, V., Sandler, M., Howard, A., & Le, Q. V. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00293 · Cited by 3,416 · cited 1×
“The recent developing trends in object detection is designing light-weight frameworks for mobile applications [5,44,48], which usually requires real-time, low-power and fully embedded.”
BackgroundSelf-citationFitNets: Hints for Thin Deep Nets
Romero, A., Ballas, N., Kahou, S. E., Chassang, A., Gatta, C., & Bengio, Y. (2014) · arXiv · doi:10.48550/arxiv.1412.6550 · Cited by 2,859
Zagoruyko, S., & Komodakis, N. (2016) · arXiv · doi:10.48550/arxiv.1612.03928 · Cited by 2,027
Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation
Bengio, Y., Léonard, N., & Courville, A. (2013) · arXiv · doi:10.48550/arxiv.1308.3432 · Cited by 1,910 · cited 1×
“The most commonly used approach is the straight-through estimator (STE) [2].”
MethodDoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients
Zhou, S., Wu, Y., Ni, Z., Zhou, X., Wen, H., & Zou, Y. (2016) · arXiv · doi:10.48550/arxiv.1606.06160 · Cited by 1,690
Hubara, I., Soudry, D., & Yaniv, R. E. (2016) · arXiv · doi:10.48550/arxiv.1602.02505 · Cited by 1,667
Deep Learning with Low Precision by Half-Wave Gaussian Quantization
Cai, Z., He, X., Sun, J., & Vasconcelos, N. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.574 · Cited by 498 · cited 1×
“To reduce the quantization error, non-uniform strategies [4, 52] propose to learn the quantization intervals by jointly optimizing parameters and quantizers.”
MethodTowards Accurate Binary Convolutional Neural Network
Lin, X., Zhao, C., & Pan, W. (2017) · arXiv · doi:10.48550/arxiv.1711.11294 · Cited by 453 · cited 1×
“The development of BNNs can be clas-sified into two categories: (i) a focus on improving the training of BNNs [13, 30, 36, 45]; (ii) multiple binariza-tions to approximate the full-precision tensor or structure [9, 23, 27, 28, 45, 58].”
MethodLearning to Quantize Deep Networks by Optimizing Quantization Intervals With Task Loss
Jung, S., Son, C., Lee, S., Son, J., Han, J.-J., Kwak, Y., Hwang, S. J., & Choi, C. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00448 · Cited by 385
How to Train a Compact Binary Neural Network with High Accuracy?
Tang, W., Hua, G., & Wang, L. (2017) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v31i1.10862 · Cited by 241
Fully Quantized Network for Object Detection
Li, R., Wang, Y., Liang, F., Qin, H., Yan, J., & Fan, R. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00292 · Cited by 214 · cited 3×
“Specifically, unlike [22] which freezes the batch normalization (BN) statistics during training to stabilize optimization, we instead propose an alternative strategy where the BN statistics still keeps updating: we propose that except the last layers for classification and regression, parameters of…”
BackgroundMethodMixed Precision Quantization of ConvNets via Differentiable Neural Architecture Search
Wu, B., Wang, Y., Zhang, P., Tian, Y., Vajda, P., & Keutzer, K. (2018) · arXiv · doi:10.48550/arxiv.1812.00090 · Cited by 186 · cited 1×
“Moreover, with the popularity of automatic machine learning, some recent literature employs reinforcement learning to search for the optimal bitwidth for each layer [6, 46, 49].”
BackgroundTwo-Step Quantization for Low-bit Neural Networks
Wang, P., Hu, Q., Zhang, Y., Zhang, C., Liu, Y., & Cheng, J. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00460 · Cited by 136 · cited 1×
“To solve this problem, some works have studied relaxed quantization [1, 31, 47, 57].”
BackgroundNetwork Sketching: Exploiting Binary Structure in Deep CNNs
Guo, Y., Yao, A., Zhao, H., & Chen, Y. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.430 · Cited by 81 · cited 1×
“The development of BNNs can be clas-sified into two categories: (i) a focus on improving the training of BNNs [13, 30, 36, 45]; (ii) multiple binariza-tions to approximate the full-precision tensor or structure [9, 23, 27, 28, 45, 58].”
MethodProxQuant: Quantized Neural Networks via Proximal Operators
Bai, Y., Wang, Y.-X., & Liberty, E. (2018) · arXiv · doi:10.48550/arxiv.1810.00861 · Cited by 76
Research works printed but not linked: 35
- redmon, 2018, YOLOv3 An Incremental Improvement
- 10.1109/CVPR.2017.690 · DOI · Paper page
- 10.1109/CVPR.2019.00934 · DOI · Paper page
- mishra, 0, Apprentice: Using knowledge distillation techniques to improve low-precision network accuracy, Proc Int Conf on Learning Rep
- louizos, 0, Relaxed quantization for discretized neural networks, Proc Int Conf on Learning Rep
- liu, 0, Bi-real net: Enhancing the performance of 1-bit cnns with improved representational capability and advanced training algorithm, Proc Eur Conf Comp Vis, 722
- pham, 0, Efficient neural architecture search via parameter sharing, Proc Int Conf Mach Learn, 4092
- 10.1109/CVPR.2019.00409 · DOI · Paper page
- liu, 0, Circulant binary convolutional networks: Enhancing the performance of 1-bit dcnns with circulant back propagation, Proc IEEE Conf Comp Vis Patt Recogn, 2691
- liu, 0, Ssd: Single shot multibox detector, Proc Eur Conf Comp Vis, 21
- lee, 2015, Deeply-supervised nets, Artificial Intelligence and Statistics, 562
- 10.1109/ICCV.2017.282 · DOI · Paper page
- lin, 0, Microsoft coco: Common objects in context, Proc Eur Conf Comp Vis, 740
- lin, 0, Focal loss for dense object detection, Proc IEEE Int Conf Comp Vis, 2980
- wu, 2019, Detectron 2
- 10.1109/CVPR.2019.00050 · DOI · Paper page
- 10.1109/CVPR.2018.00826 · DOI · Paper page
- zhuang, 2019, Effective training of convolutional neural networks with low-bitwidth weights and activations
- zhu, 0, Trained ternary quantization, Proc Int Conf on Learning Rep
- 10.1109/CVPR.2017.660 · DOI · Paper page
- zhang, 0, Lq-nets: Learned quantization for highly accurate and compact deep neural networks, Proc Eur Conf Comp Vis, 365
- 10.1109/ICCV.2017.322 · DOI · Paper page
- hinton, 0, Distilling the knowledge in a neural network, Proc Adv Neural Inf Process Syst Workshops
- hou, 0, Loss-aware binarization of deep networks, Proc Int Conf on Learning Rep
- 10.1007/978-3-030-01231-1_21 · DOI
- krizhevsky, 2009, Learning multiple layers of features from tiny images
- bethge, 2018, Training competitive binary neural networks from scratch
- chen, 2018, Joint neural architecture search and quantization
- chen, 0, Learning efficient object detection models with knowledge distillation, Proc Adv Neural Inf Process Syst, 742
- 10.1109/CVPR.2014.81 · DOI · Paper page
- 10.1109/ICCV.2015.169 · DOI · Paper page
- wang, 0, Haq: Hardware-aware automated quantization, Proc IEEE Conf Comp Vis Patt Recogn
- wei, 0, Quantization mimic: Towards very tiny cnn for object detection, Proc Eur Conf Comp Vis, 267
- 10.1007/s11263-015-0816-y · DOI · Paper page
- 10.1109/CVPR.2015.7298594 · DOI · Paper page
Cited by
Bringing ai to edge: from deep learning’s perspective
Liu, D., Kong, H., Luo, X., Liu, W., & Subramaniam, R. (2022) · Neurocomputing · doi:10.1016/j.neucom.2021.04.141 · Cited by 171
“Zhang et al. [125] adopt an auxiliary full-precision model to facilitate the training of its quantized counterparts.”
BackgroundMethodNeural inference at the frontier of energy, space, and time
Modha, D. S., Akopyan, F., Andreopoulos, A., Appuswamy, R., Arthur, J. V., Cassidy, A. S., Datta, P., DeBole, M. V., Esser, S. K., Otero, C. O., et al. (2023) · Science · doi:10.1126/science.adh1174 · Cited by 115
AQD: Towards Accurate Quantized Object Detection
Chen, P., Liu, J., Zhuang, B., Tan, M., & Shen, C. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.00017 · Cited by 29
Self-citation
Metadata
- Authors
- Zhuang, B., Liu, L., Tan, M., Shen, C., & Reid, I.
- Venue
- 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
- Publisher
- IEEE
- Published
- 2020-06
- Pages
- 1485-1494
- ISSN
- 1063-6919 (ISSN-L)
- OpenAlex
- W2991002474
- Cited by
- 75
- References
- 23
- FWCI
- 4.70
- Type
- conference paper
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.