Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
110 tokens/sec
GPT-4o
56 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

TVE: Learning Meta-attribution for Transferable Vision Explainer (2312.15359v2)

Published 23 Dec 2023 in cs.LG, cs.AI, and cs.CV

Abstract: Explainable machine learning significantly improves the transparency of deep neural networks. However, existing work is constrained to explaining the behavior of individual model predictions, and lacks the ability to transfer the explanation across various models and tasks. This limitation results in explaining various tasks being time- and resource-consuming. To address this problem, we introduce a Transferable Vision Explainer (TVE) that can effectively explain various vision models in downstream tasks. Specifically, the transferability of TVE is realized through a pre-training process on large-scale datasets towards learning the meta-attribution. This meta-attribution leverages the versatility of generic backbone encoders to comprehensively encode the attribution knowledge for the input instance, which enables TVE to seamlessly transfer to explain various downstream tasks, without the need for training on task-specific data. Empirical studies involve explaining three different architectures of vision models across three diverse downstream datasets. The experimental results indicate TVE is effective in explaining these tasks without the need for additional training on downstream data.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (44)
  1. Towards better understanding of gradient-based attribution methods for deep neural networks. arXiv preprint arXiv:1711.06104, 2017.
  2. Towards fair patient-trial matching via patient-criterion level fairness constraint. arXiv preprint arXiv:2303.13790, 2023.
  3. Rev: information-theoretic evaluation of free-text rationales. arXiv preprint arXiv:2210.04982, 2022.
  4. Harsanyinet: Computing accurate shapley values in a single forward propagation. arXiv preprint arXiv:2304.01811, 2023.
  5. A simple framework for contrastive learning of visual representations. In International conference on machine learning, pages 1597–1607. PMLR, 2020.
  6. Sasank Chilamkurthy. Transfer learning for computer vision tutorial. PyTorch Tutorials, 2017.
  7. Efficient xai techniques: A taxonomic survey. arXiv preprint arXiv:2302.03225, 2023.
  8. Cortx: Contrastive framework for real-time explanation. arXiv preprint arXiv:2303.02794, 2023.
  9. Learning to estimate shapley values with vision transformers. arXiv preprint arXiv:2206.05282, 2022.
  10. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee, 2009.
  11. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020.
  12. Techniques for interpretable machine learning. Communications of the ACM, 63(1):68–77, 2019.
  13. Asirra: A captcha that exploits interest-aligned manual image categorization. In Proceedings of 14th ACM Conference on Computer and Communications Security (CCS). Association for Computing Machinery, Inc., October 2007.
  14. Understanding dataset difficulty with v-usable information. pages 5988–6008, 2022.
  15. Luciano Floridi. Establishing the rules for building trustworthy ai. Nature Machine Intelligence, 1(6):261–262, 2019.
  16. European union regulations on algorithmic decision-making and a “right to explanation”. AI magazine, 38(3):50–57, 2017.
  17. Masked autoencoders are scalable vision learners. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16000–16009, 2022.
  18. Momentum contrast for unsupervised visual representation learning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9729–9738, 2020.
  19. Conditional probing: measuring usable information beyond a baseline. arXiv preprint arXiv:2109.09234, 2021.
  20. Jeremy Howard. Imagenette: A smaller subset of 10 easily classified classes from imagenet, March 2019.
  21. Fastshap: Real-time shapley value estimation. In International Conference on Learning Representations, 2021.
  22. Segment anything. arXiv preprint arXiv:2304.02643, 2023.
  23. Captum: A unified and generic model interpretability library for pytorch. arXiv preprint arXiv:2009.07896, 2020.
  24. Learning multiple layers of features from tiny images. 2009.
  25. Synthetic benchmarks for scientific research in explainable machine learning. 2021.
  26. Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF international conference on computer vision, pages 10012–10022, 2021.
  27. A unified approach to interpreting model predictions. Advances in neural information processing systems, 30, 2017.
  28. Sampling permutations for shapley value estimation. The Journal of Machine Learning Research, 23(1):2082–2127, 2022.
  29. Rise: Randomized input sampling for explanation of black-box models. arXiv preprint arXiv:1806.07421, 2018.
  30. ” why should i trust you?” explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, pages 1135–1144, 2016.
  31. Efficient gnn explanation via learning removal-based attribution. arXiv preprint arXiv:2306.05760, 2023.
  32. Grad-cam: Visual explanations from deep networks via gradient-based localization. In Proceedings of the IEEE international conference on computer vision, pages 618–626, 2017.
  33. On the web’s cutting edge, anonymity in name only. The Wall Street Journal, 4, 2010.
  34. Explain any concept: Segment anything meets concept-based explanation. arXiv preprint arXiv:2305.10289, 2023.
  35. Singular value fine-tuning: Few-shot segmentation requires few-parameters fine-tuning. Advances in Neural Information Processing Systems, 35:37484–37496, 2022.
  36. Axiomatic attribution for deep networks. In International conference on machine learning, pages 3319–3328. PMLR, 2017.
  37. Training data-efficient image transformers & distillation through attention. In International conference on machine learning, pages 10347–10357. PMLR, 2021.
  38. Accelerating shapley explanation via contributive cooperator selection. arXiv preprint arXiv:2206.08529, 2022.
  39. Transformers: State-of-the-art natural language processing. In Proceedings of the 2020 conference on empirical methods in natural language processing: system demonstrations, pages 38–45, 2020.
  40. Transformers: State-of-the-art natural language processing. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, pages 38–45, Online, October 2020. Association for Computational Linguistics.
  41. A theory of usable information under computational constraints. arXiv preprint arXiv:2002.10689, 2020.
  42. Towards interpretation of recommender systems with sorted explanation paths. In 2018 IEEE International Conference on Data Mining (ICDM), pages 667–676. IEEE, 2018.
  43. Tutorial on deep learning interpretation: A data perspective. In Proceedings of the 31st ACM International Conference on Information & Knowledge Management, pages 5156–5159, 2022.
  44. Anna Zimdars. Fairness and undergraduate admission: a qualitative exploration of admissions choices at the university of oxford. Oxford Review of Education, 36(3):307–323, 2010.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (11)
  1. Guanchu Wang (33 papers)
  2. Yu-Neng Chuang (28 papers)
  3. Fan Yang (878 papers)
  4. Mengnan Du (90 papers)
  5. Chia-Yuan Chang (18 papers)
  6. Shaochen Zhong (15 papers)
  7. Zirui Liu (58 papers)
  8. Zhaozhuo Xu (43 papers)
  9. Kaixiong Zhou (52 papers)
  10. Xuanting Cai (13 papers)
  11. Xia Hu (186 papers)
Citations (1)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now