mlsanityData to 14 Oct 2026

Comprehensive Information Bottleneck for Unveiling Universal Attribution to Interpret Vision Transformers

arXiv 2507.04388 ↗First posted 6 Jul 2025cs.CVcited by 6

Not on the map. Why?

Abstract (arXiv)

The feature attribution method reveals the contribution of input variables to the decision-making process to provide an attribution map for explanation. Existing methods grounded on the information bottleneck principle compute information in a specific layer to obtain attributions, compressing the features by injecting noise via a parametric damping ratio. However, the attribution obtained in a specific layer neglects evidence of the decision-making process distributed across layers. In this paper, we introduce a comprehensive information bottleneck (CoIBA), which discovers the relevant information in each targeted layer to explain the decision-making process. Our core idea is applying information bottleneck in multiple targeted layers to estimate the comprehensive information by sharing a parametric damping ratio across the layers. Leveraging this shared ratio complements the over-compressed information to discover the omitted clues of the decision by sharing the relevant information across the targeted layers. We suggest the variational approach to fairly reflect the relevant information of each layer by upper bounding layer-wise information. Therefore, CoIBA guarantees that the discarded activation is unnecessary in every targeted layer to make a decision. The extensive experimental results demonstrate the enhancement in faithfulness of the feature attributions provided by CoIBA.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 33 of its 43 references that are in the map. First posted on arXiv 6 Jul 2025.

Builds on

33 of 43 references.

References · 33

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  4. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,403
  5. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,968
  6. 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,733
  7. 2020Training data-efficient image transformers & distillation through attentioncited by 4,586
  8. 2017Axiomatic Attribution for Deep Networkscited by 3,384
  9. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,109
  10. 2021BEiT: BERT Pre-Training of Image Transformerscited by 2,375
  11. 2020The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalizationcited by 1,850
  12. 2019Similarity of Neural Network Representations Revisitedcited by 1,755
21 more
  1. 2017Towards A Rigorous Science of Interpretable Machine Learningcited by 1,416
  2. 2019Natural Adversarial Examplescited by 1,408
  3. 2021Swin Transformer V2: Scaling Up Capacity and Resolutioncited by 1,228
  4. 2018Sanity Checks for Saliency Mapscited by 1,156
  5. 2016Deep Variational Information Bottleneckcited by 930
  6. 2018RISE: Randomized Input Sampling for Explanation of Black-box Modelscited by 838
  7. 2020Quantifying Attention Flow in Transformerscited by 775
  8. 2022EVA: Exploring the Limits of Masked Visual Representation Learning at Scalecited by 691
  9. 2020Transformer Interpretability Beyond Attention Visualizationcited by 627
  10. 2021How to train your ViT? Data, Augmentation, and Regularization in Vision Transformerscited by 569
  11. 2018A Benchmark for Interpretability Methods in Deep Neural Networkscited by 476
  12. 2022DeiT III: Revenge of the ViTcited by 419
  13. 2021Generic Attention-model Explainability for Interpreting Bi-Modal and Encoder-Decoder Transformerscited by 320
  14. 2016Layer-wise Relevance Propagation for Neural Networks with Local Renormalization Layerscited by 199
  15. 2020Restricting the Flow: Information Bottlenecks for Attributioncited by 110
  16. 2022A Consistent and Efficient Evaluation Strategy for Attribution Methodscited by 102
  17. 2023FunnyBirds: A Synthetic Vision Dataset for a Part-Based Analysis of Explainable AI Methodscited by 50
  18. 2022Optimizing Relevance Maps of Vision Transformers Improves Robustnesscited by 41
  19. 2022ViT-CX: Causal Explanation of Vision Transformerscited by 28
  20. 2023Visual Explanations via Iterated Integrated Attributionscited by 19
  21. 2021Fine-Grained Neural Network Explanation by Identifying Input Features with Predictive Informationcited by 17

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 5.

6 papers.

  1. 2025Distillation Dynamics: Towards Understanding Feature-Based Distillation in Vision Transformerscited by 2
  2. 2026From independent patches to coordinated attention: Controlling information flow in vision transformerscited by 1
  3. 2026GeoIB: Geometry-Aware Information Bottleneck via Statistical-Manifold Compressioncited by 0
  4. 2026Attribution via Distributional Paths for Information Revelationcited by 0
  5. 2026IB-HFN: Information Bottleneck-Driven SAR-Optical Fusion Network for High-Fidelity Cloud Removalcited by 0
  6. 2026Information-Regularized Attention for Visual-Centric Reasoningcited by 0