Multi-Dimensional Vision Transformer Compression via Dependency Guided Gaussian Process Search
Abstract
Vision transformers (ViT) have recently attracted considerable attentions, but the huge computational cost remains an issue for practical deployment. Previous ViT pruning methods tend to prune the model along one dimension solely, which may suffer from excessive reduction and lead to sub-optimal model quality. In contrast, we advocate a multi-dimensional ViT compression paradigm, and propose to harness the redundancy reduction from attention head, neuron and sequence dimensions jointly. Firstly, we propose a statistical dependence based pruning criterion that is generalizable to different dimensions for identifying the deleterious components. Moreover, we cast the multidimensional ViT compression as an optimization problem, objective of which is to learn an optimal pruning policy across the three dimensions while maximizing the compressed model’s accuracy under a computational budget. The problem is solved by an adapted Gaussian process search with expected improvement. Experimental results show that our method effectively reduces the computational cost of various ViT models. For example, our method reduces 40% FLOPs without top-1 accuracy loss for DeiT and T2T-ViT models on the ImageNet dataset, outperforming previous state-of-the-art ViT pruning methods.
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?
Vision Transformers suffer from higher memory footprints, computational overheads, and power consumption than convolutional neural networks, hindering their deployment and real-time inference on resource-constrained edge devices.
“Nevertheless, as compared to convolutional neural networks (CNNs), ViTs suffer from higher memory footprints, computational overheads, and power consumption, hindering their deployment and real-time inference on resourceconstrained edge devices [25, 17, 15, 37].”
From I-ViT: Integer-only Quantization for Efficient Vision Transformer Inference (2023) · cites this paper as [17]
Cites
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2020) · arXiv · doi:10.48550/arxiv.2010.11929 · Cited by 21,793 · see citing works
Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions
Wang, W., Xie, E., Li, X., Fan, D.-P., Song, K., Liang, D., Lu, T., Luo, P., & Shao, L. (2021) · 2021 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00061 · Cited by 5,259 · cited 5× · see citing works
“Accordingly, we modify our sequence reduction strategy for PVT-Small.”
MethodInfluentialLinformer: Self-Attention with Linear Complexity
Wang, S., Li, B. Z., Khabsa, M., Fang, H., & Ma, H. (2020) · arXiv · doi:10.48550/arxiv.2006.04768 · Cited by 867 · see citing works
DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification
Rao, Y., Zhao, W., Liu, B., Lu, J., Zhou, J., & Hsieh, C.-J. (2021) · arXiv · doi:10.48550/arxiv.2106.02034 · Cited by 448 · cited 6× · see citing works
“We compare with the latest ViT model compression methods, including sequence reduction methods (DynamicViT [23], IA-RED(2) [22], PatchSlim [25], Evo-ViT [33]), weight pruning methods (VTP [38], S(2)ViTE [5]), unified ViT compression method UVC [36], and NAS-based ViT pruning method [12].”
BackgroundMethodInfluentialNyströmformer: A Nyström-based Algorithm for Approximating Self-Attention
Xiong, Y., Zeng, Z., Chakraborty, R., Tan, M., Fung, G., Li, Y., & Singh, V. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i16.17664 · Cited by 376 · see citing works
AutoFormer: Searching Transformers for Visual Recognition
Chen, M., Peng, H., Fu, J., & Ling, H. (2021) · 2021 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.01205 · Cited by 323 · see citing works
Patch Slimming for Efficient Vision Transformers
Tang, Y., Han, K., Wang, Y., Xu, C., Guo, J., Xu, C., & Tao, D. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.01185 · Cited by 201 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
I-ViT: Integer-only Quantization for Efficient Vision Transformer Inference
Li, Z., & Gu, Q. (2023) · 2023 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01565 · Cited by 141 · see citing works
“Nevertheless, as compared to convolutional neural networks (CNNs), ViTs suffer from higher memory footprints, computational overheads, and power consumption, hindering their deployment and real-time inference on resourceconstrained edge devices [16,18,25,35].”
BackgroundRepQ-ViT: Scale Reparameterization for Post-Training Quantization of Vision Transformers
Li, Z., Xiao, J., Yang, L., & Gu, Q. (2023) · 2023 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01580 · Cited by 127 · see citing works
- Authors
- Hou, Z., & Kung, S.-Y.
- Affiliations
- Princeton University (US)
- Venue
- 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
- Publisher
- IEEE
- Published
- 2022-06
- Pages
- 3668-3677
- OpenAlex
- W4292794054
- Cited by
- 20
- References
- 7
- Citations by year
- 2026: 2 · 2025: 9 · 2024: 3 · 2023: 6
- FWCI
- 1.63
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Keywords
- attention head pruning, neuron pruning, expected improvement, neural architecture search, model compression, FLOPs reduction
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
