Ternary MobileNets via Per-Layer Hybrid Filter Banks
Open accessThe DOI matches its Crossref record; 20 of the 20 references it cites were found in a registry or an index.
Abstract
MobileNets family of computer vision neural networks have fueled tremendous progress in the design and organization of resource-efficient architectures in recent years. New applications with stringent real-time requirements on highly constrained devices require further compression of MobileNets-like compute-efficient networks. Model quantization is a widely used technique to compress and accelerate neural network inference and prior works have quantized MobileNets to 4 - 6 bits, albeit with a modest to significant drop in accuracy. While quantization to sub-byte values (i.e. precision ≤ 8 bits) has been valuable, even further quantization of MobileNets to binary or ternary values is necessary to realize significant energy savings and possibly runtime speedups on specialized hardware, such as ASICs and FPGAs. Under the key observation that convolutional filters at each layer of a deep neural network may respond differently to ternary quantization, we propose a novel quantization method that generates per-layer hybrid filter banks consisting of full-precision and ternary weight filters for MobileNets. Using this proposed quantization method, we quantize a substantial portion of weight filters of MobileNets to ternary values resulting in a 27.98% savings in energy, and a 51.07% reduction in the model size, while achieving comparable accuracy and no degradation in throughput on specialized hardware in comparison to the baseline full-precision MobileNets. Finally, we demonstrate the generalizability and effectiveness of hybrid filter banks to other neural network architectures.
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: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333
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
Channel Pruning for Accelerating Very Deep Neural Networks
He, Y., Zhang, X., & Sun, J. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.155 · Cited by 2,647 · cited 1×
“As different channel pruning [19, 54, 20] and filter pruning techniques [18, 32, 51, 16, 1, 42, 30, 45, 15] are orthogonal to our compression scheme, they can be used in conjunction with hybrid filter banks to further reduce model size and computational complexity.”
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
AMC: AutoML for Model Compression and Acceleration on Mobile Devices
He, Y., Lin, J., Liu, Z., Wang, H., Li, L.-J., & Han, S. (2018) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-01234-2_48 · Cited by 1,400
Speeding up Convolutional Neural Networks with Low Rank Expansions
Jaderberg, M., Vedaldi, A., & Zisserman, A. (2014) · Proceedings of the British Machine Vision Conference 2014 · doi:10.5244/c.28.88 · Cited by 1,164
HAQ: Hardware-Aware Automated Quantization With Mixed Precision
Wang, K., Liu, Z., Lin, Y., Lin, J., & Han, S. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00881 · Cited by 978
To prune, or not to prune: exploring the efficacy of pruning for model compression
Zhu, M., & Gupta, S. (2017) · arXiv · doi:10.48550/arxiv.1710.01878 · Cited by 805 · cited 1×
“As different channel pruning [19, 54, 20] and filter pruning techniques [18, 32, 51, 16, 1, 42, 30, 45, 15] are orthogonal to our compression scheme, they can be used in conjunction with hybrid filter banks to further reduce model size and computational complexity.”
MethodLi, F., Liu, B., Wang, X., Zhang, B., & Yan, J. (2016) · arXiv · doi:10.48550/arxiv.1605.04711 · Cited by 732
NetAdapt: Platform-Aware Neural Network Adaptation for Mobile Applications
Yang, T.-J., Howard, A., Chen, B., Zhang, X., Go, A., Sandler, M., Sze, V., & Adam, H. (2018) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-01249-6_18 · Cited by 572
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×
“In order to observe the impact of ternary quantization [10, 33, 28, 5, 27, 50, 49], we apply the ternary weight quantization method from [27] over the baseline MobileNets-V1 architecture.”
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×
“In order to observe the impact of ternary quantization [10, 33, 28, 5, 27, 50, 49], we apply the ternary weight quantization method from [27] over the baseline MobileNets-V1 architecture.”
MethodLow-bit Quantization of Neural Networks for Efficient Inference
Choukroun, Y., Kravchik, E., Yang, F., & Kisilev, P. (2019) · 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW) · doi:10.1109/iccvw.2019.00363 · Cited by 404 · cited 1×
“There are recent works [41, 4, 9] that can quantize MobileNets with 4-6-bit weights (see Appendix and Table 2 in [29] for more details).”
BackgroundLearning 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
YodaNN: An Architecture for Ultralow Power Binary-Weight CNN Acceleration
Andri, R., Cavigelli, L., Rossi, D., & Benini, L. (2018) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2017.2682138 · Cited by 247
Binary Ensemble Neural Network: More Bits per Network or More Networks per Bit?
Zhu, S., Dong, X., & Su, H. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00506 · Cited by 153 · cited 1×
“In recent years, numerous research efforts have been devoted to quantizing ResNet architecture to ternary values while preserving the accuracy of full-precision model [40, 29, 44, 11, 26, 53, 52, 35, 46, 17].”
BackgroundPost-training 4-bit quantization of convolution networks for rapid-deployment
Banner, R., Nahshan, Y., Hoffer, E., & Soudry, D. (2018) · arXiv · doi:10.48550/arxiv.1810.05723 · Cited by 115
Network 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×
“In recent years, numerous research efforts have been devoted to quantizing ResNet architecture to ternary values while preserving the accuracy of full-precision model [40, 29, 44, 11, 26, 53, 52, 35, 46, 17].”
BackgroundCompressing RNNs for IoT devices by 15-38x using Kronecker Products
Thakker, U., Beu, J., Gope, D., Zhou, C., Fedorov, I., Dasika, G., & Mattina, M. (2019) · arXiv · doi:10.48550/arxiv.1906.02876 · Cited by 30 · cited 1×
“Tensor decomposition techniques [24, 36, 43, 39, 37, 38] exploit parameter redundancy to obtain low-rank approximations of weight matrices.”
MethodSelf-citation
Cited by
Banbury, C., Zhou, C., Fedorov, I., Navarro, R. M., Thakker, U., Gope, D., Reddi, V. J., Mattina, M., & Whatmough, P. N. (2020) · arXiv · doi:10.48550/arxiv.2010.11267 · Cited by 183
Self-citationFederated Learning for Resource-Constrained IoT Devices: Panoramas and State of the Art
Imteaj, A., Mamun Ahmed, K., Thakker, U., Wang, S., Li, J., & Amini, M. H. (2023) · Adaptation, Learning, and Optimization · doi:10.1007/978-3-031-11748-0_2 · Cited by 86
Self-citationCompressing RNNs to Kilobyte Budget for IoT Devices Using Kronecker Products
Thakker, U., Fedorov, I., Zhou, C., Gope, D., Mattina, M., Dasika, G., & Beu, J. (2021) · ACM Journal on Emerging Technologies in Computing Systems · doi:10.1145/3440016 · Cited by 8
Self-citationCompressing Language Models using Doped Kronecker Products
Thakker, U., Whatmough, P. N., Liu, Z.-G., Mattina, M., & Beu, J. (2020) · arXiv · doi:10.48550/arxiv.2001.08896 · Cited by 7
“…categorized under 4 topics - Pruning (Han et al., 2016; Zhu & Gupta, 2017), structured matrix based techniques (Sindhwani et al., 2015; Thakker et al., 2019b;d), quantization (Hubara et al., 2016; Courbariaux & Bengio, 2016; Gope et al., 2019) and tensor decomposition (Tjandra et al., 2017).”
Self-citation
Metadata
- Authors
- Gope, D., Beu, J., Thakker, U., & Mattina, M.
- Venue
- 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
- Publisher
- IEEE
- Published
- 2020-06
- Pages
- 3036-3046
- OpenAlex
- W2984074432
- Cited by
- 15
- References
- 20
- FWCI
- 1.05
- 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.