Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
167 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
42 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Learning Decomposable and Debiased Representations via Attribute-Centric Information Bottlenecks (2403.14140v1)

Published 21 Mar 2024 in cs.CV and cs.LG

Abstract: Biased attributes, spuriously correlated with target labels in a dataset, can problematically lead to neural networks that learn improper shortcuts for classifications and limit their capabilities for out-of-distribution (OOD) generalization. Although many debiasing approaches have been proposed to ensure correct predictions from biased datasets, few studies have considered learning latent embedding consisting of intrinsic and biased attributes that contribute to improved performance and explain how the model pays attention to attributes. In this paper, we propose a novel debiasing framework, Debiasing Global Workspace, introducing attention-based information bottlenecks for learning compositional representations of attributes without defining specific bias types. Based on our observation that learning shape-centric representation helps robust performance on OOD datasets, we adopt those abilities to learn robust and generalizable representations of decomposable latent embeddings corresponding to intrinsic and biasing attributes. We conduct comprehensive evaluations on biased datasets, along with both quantitative and qualitative analyses, to showcase our approach's efficacy in attribute-centric representation learning and its ability to differentiate between intrinsic and bias-related features.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (94)
  1. Towards causal vqa: Revealing and reducing spurious correlations by invariant and covariant semantic editing. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 9690–9698.
  2. Adversarial invariant feature learning with accuracy constraint for domain generalization. In Proceedings of the Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD, Part II, pages 315–331. Springer.
  3. Towards robust interpretability with self-explaining neural networks. Advances in Neural Information Processing Systems, 31.
  4. Layer normalization. arXiv preprint arXiv:1607.06450.
  5. Baars, B. J. (1993). A cognitive theory of consciousness. Cambridge University Press.
  6. Learning de-biased representations with biased representations. In Proceedings of the International Conference on Machine Learning, pages 528–539. PMLR.
  7. Entropy-based logic explanations of neural networks. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 6046–6054.
  8. Monet: Unsupervised scene decomposition and representation. arXiv preprint arXiv:1901.11390.
  9. Object representations as fixed points: Training iterative refinement algorithms with implicit differentiation. Advances in Neural Information Processing Systems, 35:32694–32708.
  10. This looks like that: deep learning for interpretable image recognition. Advances in Neural Information Processing Systems, 32.
  11. Concept whitening for interpretable image recognition. Nature Machine Intelligence, 2(12):772–782.
  12. Learning phrase representations using RNN encoder–decoder for statistical machine translation. In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1724–1734.
  13. Algorithms for learning kernels based on centered alignment. The Journal of Machine Learning Research, 13(1):795–828.
  14. Latent adversarial debiasing: Mitigating collider bias in deep neural networks. arXiv preprint arXiv:2011.11486.
  15. Experimental and theoretical approaches to conscious processing. Neuron, 70(2):200–227.
  16. Imagenet: A large-scale hierarchical image database. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee.
  17. Improving interpretability of deep neural networks with semantic information. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4306–4314.
  18. An image is worth 16x16 words: Transformers for image recognition at scale. In International Conference on Learning Representations.
  19. Connectionism and cognitive architecture: A critical analysis. Cognition, 28(1-2):3–71.
  20. Domain-adversarial training of neural networks. Journal of machine learning research, 17(59):1–35.
  21. Shortcut learning in deep neural networks. Nature Machine Intelligence, 2(11):665–673.
  22. Partial success in closing the gap between human and machine vision. Advances in Neural Information Processing Systems, 34:23885–23899.
  23. Imagenet-trained cnns are biased towards texture; increasing shape bias improves accuracy and robustness. In Proceedings of the International Conference on Learning Representations.
  24. Towards automatic concept-based explanations. Advances in Neural Information Processing Systems, 32.
  25. Model patching: Closing the subgroup performance gap with data augmentation. In Proceedings of the International Conference on Learning Representations.
  26. Inductive biases for deep learning of higher-level cognition. Proceedings of the Royal Society A, 478(2266):20210068.
  27. Coordination among neural modules through a shared global workspace. In Proceedings of the International Conference on Learning Representations.
  28. Recurrent independent mechanisms. arXiv preprint arXiv:1909.10893.
  29. Explaining classifiers with causal concept effect (CaCE). arXiv preprint arXiv:1907.07165.
  30. Multi-object representation learning with iterative variational inference. In International conference on machine learning, pages 2424–2433. PMLR.
  31. On the binding problem in artificial neural networks. arXiv preprint arXiv:2012.05208.
  32. On calibration of modern neural networks. In Proceedings of the International Conference on Machine Learning (ICML), pages 1321–1330.
  33. An analysis of explainability methods for convolutional neural networks. Engineering Applications of Artificial Intelligence, 117:105606.
  34. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778.
  35. Bag of tricks for image classification with convolutional neural networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 558–567.
  36. Benchmarking neural network robustness to common corruptions and perturbations. In International Conference on Learning Representations.
  37. Concept-centric transformers: Enhancing model interpretability through object-centric concept learning within a shared global workspace. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pages 4880–4891.
  38. Arbitrary style transfer in real-time with adaptive instance normalization. In Proceedings of the IEEE international conference on computer vision, pages 1501–1510.
  39. Perceiver: General perception with iterative attention. In Proceedings of the International Conference on Machine Learning, pages 4651–4664. PMLR.
  40. Improving object-centric learning with query optimization. In Proceedings of the International Conference on Learning Representations.
  41. A style-based generator architecture for generative adversarial networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4401–4410.
  42. Image generation using generative adversarial networks and attention mechanism. In Proceedings of the IEEE/ACIS International Conference on Computer and Information Science (ICIS), pages 1–6. IEEE.
  43. Now you see me (CME): concept-based model extraction. arXiv preprint arXiv:2010.13233.
  44. Learning not to learn: Training deep neural networks with biased data. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9012–9020.
  45. Interpretability beyond feature attribution: Quantitative testing with concept activation vectors (TCAV). In Proceedings of the International Conference on Machine Learning, pages 2668–2677. PMLR.
  46. Biaswap: Removing dataset bias with bias-tailored swapping augmentation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 14992–15001.
  47. The (un) reliability of saliency methods. Explainable AI: Interpreting, explaining and visualizing deep learning, pages 267–280.
  48. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980.
  49. Concept bottleneck models. In Proceedings of the International Conference on Machine Learning, pages 5338–5348. PMLR.
  50. Similarity of neural network representations revisited. In Proceedings of the International Conference on Machine Learning, pages 3519–3529. PMLR.
  51. Learning multiple layers of features from tiny images.(2009).
  52. Learning debiased representation via disentangled feature augmentation. Advances in Neural Information Processing Systems, 34:25123–25133.
  53. Deep learning for case-based reasoning through prototypes: A neural network that explains its predictions. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 32.
  54. Repair: Removing representation bias by dataset resampling. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9572–9581.
  55. Shape-texture debiased neural network training. arXiv preprint arXiv:2010.05981.
  56. Biasadv: Bias-adversarial augmentation for model debiasing. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3832–3841.
  57. Just train twice: Improving group robustness without training group information. In International Conference on Machine Learning, pages 6781–6792. PMLR.
  58. Object-centric learning with slot attention. Advances in Neural Information Processing Systems, 33:11525–11538.
  59. Deep dual learning for semantic image segmentation. In Proceedings of the IEEE international conference on computer vision, pages 2718–2726.
  60. Conscious processing and the global neuronal workspace hypothesis. Neuron, 105(5):776–798.
  61. Automatic shortcut removal for self-supervised representation learning. In Proceedings of the International Conference on Machine Learning, pages 6927–6937. PMLR.
  62. Minsky, M. (1988). Society of mind. Simon and Schuster.
  63. Metalearned neural memory. Advances in Neural Information Processing Systems, 32.
  64. Learning from failure: De-biasing classifier from biased classifier. Advances in Neural Information Processing Systems, 33:20673–20684.
  65. Do vision transformers see like convolutional neural networks? Advances in Neural Information Processing Systems, 34:12116–12128.
  66. Robbins, P. (2017). Modularity of mind. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, Winter 2017 edition.
  67. V-measure: A conditional entropy-based external cluster evaluation measure. In Proceedings of the Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pages 410–420.
  68. Rudin, C. (2019). Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature Machine Intelligence, 1(5):206–215.
  69. Learning from others’ mistakes: Avoiding dataset biases without modeling them. In International Conference on Learning Representations.
  70. Relational recurrent neural networks. Advances in neural information processing systems, 31.
  71. Toward causal representation learning. Proceedings of the IEEE, 109(5):612–634.
  72. Which shortcut cues will dnns choose? a study from the parameter-space perspective. In International Conference on Learning Representations.
  73. Theories of consciousness. Nature Reviews Neuroscience, 23(7):439–452.
  74. The pitfalls of simplicity bias in neural networks. Advances in Neural Information Processing Systems, 33:9573–9585.
  75. Improving shape awareness and interpretability in deep networks using geometric moments. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4158–4167.
  76. One pixel attack for fooling deep neural networks. IEEE Transactions on Evolutionary Computation, 23(5):828–841.
  77. End: Entangling and disentangling deep representations for bias correction. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 13508–13517.
  78. Unbiased look at dataset bias. In CVPR 2011, pages 1521–1528. IEEE.
  79. Multiattribute sample learning for hyperspectral image classification using hierarchical peak attribute propagation. IEEE Transactions on Instrumentation and Measurement, 71:1–17.
  80. Visualizing data using t-sne. Journal of Machine Learning Research, 9(86):2579–2605.
  81. Attention is all you need. Advances in Neural Information Processing Systems, 30.
  82. Manifold mixup: Better representations by interpolating hidden states. In International conference on machine learning, pages 6438–6447. PMLR.
  83. Learning robust representations by projecting superficial statistics out. In Proceedings of the International Conference on Learning Representations.
  84. Generative image modeling using style and structure adversarial networks. In Proceedings of the European Conference on Computer Vision, pages 318–335. Springer.
  85. Towards fairness in visual recognition: Effective strategies for bias mitigation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8919–8928.
  86. Self-training with noisy student improves imagenet classification. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10687–10698.
  87. On completeness-aware concept-based explanations in deep neural networks. Advances in Neural Information Processing Systems, 33:20554–20565.
  88. Wide residual networks. In British Machine Vision Conference 2016. British Machine Vision Association.
  89. Concept embedding models. In Proceedings of the Conference on Neural Information Processing Systems.
  90. mixup: Beyond empirical risk minimization. In International Conference on Learning Representations.
  91. Learning debiased representations via conditional attribute interpolation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 7599–7608.
  92. Generalized cross entropy loss for training deep neural networks with noisy labels. Advances in neural information processing systems, 31.
  93. Attribute hierarchy based multi-task learning for fine-grained image classification. Neurocomputing, 395:150–159.
  94. Dense semantic image segmentation with objects and attributes. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 3214–3221.

Summary

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

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

Tweets