A Comprehensive Overhaul of Feature Distillation
Open accessAbstract
We investigate the design aspects of feature distillation methods achieving network compression and propose a novel feature distillation method in which the distillation loss is designed to make a synergy among various aspects: teacher transform, student transform, distillation feature position and distance function. Our proposed distillation loss includes a feature transform with a newly designed margin ReLU, a new distillation feature position, and a partial L2distance function to skip redundant information giving adverse effects to the compression of student. In ImageNet, our proposed method achieves 21.65% of top-1 error with ResNet50, which outperforms the performance of the teacher network, ResNet152. Our proposed method is evaluated on various tasks such as image classification, object detection and semantic segmentation and achieves a significant performance improvement in all tasks. The code is available at bhheo.github.io/overhaul.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Feature maps and logits of a network are widely used as knowledge in knowledge distillation.
“Feature maps [8, 9, 10] and logits of a network [11, 12] are widely used as knowledge.”
From PQK: Model Compression via Pruning, Quantization, and Knowledge Distillation (2021) · cites this paper as [9]
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 · see citing works
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 · see citing works
ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380 · see citing works
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
Ioffe, S., & Szegedy, C. (2015) · arXiv · doi:10.48550/arxiv.1502.03167 · Cited by 27,818 · see citing works
Zagoruyko, S., & Komodakis, N. (2016) · Procedings of the British Machine Vision Conference 2016 · doi:10.5244/c.30.87 · Cited by 7,244 · see citing works
The Pascal Visual Object Classes Challenge: A Retrospective
Everingham, M., Eslami, S. M. A., Van Gool, L., Williams, C. K. I., Winn, J., & Zisserman, A. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-014-0733-5 · Cited by 7,139 · see citing works
NSML: Meet the MLaaS platform with a real-world case study
Kim, H., Kim, M., Seo, D., Kim, J., Park, H., Park, S., Jo, H., Kim, K., Yang, Y., Kim, Y., et al. (2018) · arXiv · doi:10.48550/arxiv.1810.09957 · Cited by 91 · see citing works
Self-citation
Flagged first, then by citations
Cited by
Sorted by citations
Knowledge Distillation: A Survey
Gou, J., Yu, B., Maybank, S. J., & Tao, D. (2021) · International Journal of Computer Vision · doi:10.1007/s11263-021-01453-z · Cited by 3,931 · see citing works
Decoupled Multimodal Distilling for Emotion Recognition
Li, Y., Wang, Y., & Cui, Z. (2023) · 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52729.2023.00641 · Cited by 249 · see citing works
A survey on optimization techniques for edge artificial intelligence (ai)
Surianarayanan, C., Lawrence, J. J., Chelliah, P. R., Prakash, E., & Hewage, C. (2023) · Sensors · doi:10.3390/s23031279 · Cited by 131 · see citing works
PQK: Model Compression via Pruning, Quantization, and Knowledge Distillation
Kim, J., Chang, S., & Kwak, N. (2021) · Interspeech 2021 · doi:10.21437/interspeech.2021-248 · Cited by 85 · see citing works
“Feature maps [8, 9, 10] and logits of a network [11, 12] are widely used as knowledge.”
BackgroundSelf-citationA survey of model compression strategies for object detection
Lyu, Z., Yu, T., Pan, F., Zhang, Y., Luo, J., Zhang, D., Chen, Y., Zhang, B., & Li, G. (2023) · Multimedia Tools and Applications · doi:10.1007/s11042-023-17192-x · Cited by 65 · see citing works
Knowledge from the Original Network: Restore a Better Pruned Network with Knowledge Distillation
Chen, L., Chen, Y., Xi, J., & Le, X. (2022) · Complex & Intelligent Systems · doi:10.1007/s40747-020-00248-y · Cited by 48 · see citing works
- Authors
- Heo, B., Kim, J., Yun, S., Park, H., Kwak, N., & Choi, J. Y.
- Affiliations
- Seoul National University (KR)
- Naver (South Korea) (KR)
- Venue
- 2019 IEEE/CVF International Conference on Computer Vision (ICCV)
- Publisher
- IEEE
- Published
- 2019-10
- Pages
- 1921-1930
- OpenAlex
- W2982157312
- Cited by
- 706
- References
- 7
- Citations by year
- 2026: 69 · 2025: 146 · 2024: 160 · 2023: 135 · 2022: 77 · 2021: 77 · 2020: 39 · 2019: 2 · 2018: 1
- FWCI
- 15.89
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE/CVF International Conference on Computer Vision (ICCV) (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Seoul National University Open Repository (Seoul National University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Ministry of Science, ICT and Future Planning (2017‐0‐00306, 2017M3C4A7077582)
- Institute for Information and Communications Technology Promotion (2017-0-00306)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
- Keywords
- feature distillation, knowledge distillation, network compression, teacher-student models, image classification, object detection, semantic segmentation
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
