Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
119 tokens/sec
GPT-4o
56 tokens/sec
Gemini 2.5 Pro Pro
43 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

Unveiling and Mitigating Generalized Biases of DNNs through the Intrinsic Dimensions of Perceptual Manifolds (2404.13859v3)

Published 22 Apr 2024 in cs.CV and cs.AI

Abstract: Building fair deep neural networks (DNNs) is a crucial step towards achieving trustworthy artificial intelligence. Delving into deeper factors that affect the fairness of DNNs is paramount and serves as the foundation for mitigating model biases. However, current methods are limited in accurately predicting DNN biases, relying solely on the number of training samples and lacking more precise measurement tools. Here, we establish a geometric perspective for analyzing the fairness of DNNs, comprehensively exploring how DNNs internally shape the intrinsic geometric characteristics of datasets-the intrinsic dimensions (IDs) of perceptual manifolds, and the impact of IDs on the fairness of DNNs. Based on multiple findings, we propose Intrinsic Dimension Regularization (IDR), which enhances the fairness and performance of models by promoting the learning of concise and ID-balanced class perceptual manifolds. In various image recognition benchmark tests, IDR significantly mitigates model bias while improving its performance.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (42)
  1. K. Bi, L. Xie, H. Zhang, X. Chen, X. Gu, and Q. Tian, “Accurate medium-range global weather forecasting with 3d neural networks,” Nature, vol. 619, no. 7970, pp. 533–538, 2023.
  2. C. Wang, Z. Yu, S. McAleer, T. Yu, and Y. Yang, “Asp: Learn a universal neural solver!” IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 1–13, 2024.
  3. P. Xu, X. Zhu, and D. A. Clifton, “Multimodal learning with transformers: A survey,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 10, pp. 12 113–12 132, 2023.
  4. J. Li, D. Li, S. Savarese, and S. Hoi, “Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models,” arXiv preprint arXiv:2301.12597, 2023.
  5. A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark et al., “Learning transferable visual models from natural language supervision,” in International conference on machine learning.   PMLR, 2021, pp. 8748–8763.
  6. F.-A. Croitoru, V. Hondru, R. T. Ionescu, and M. Shah, “Diffusion models in vision: A survey,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 9, pp. 10 850–10 869, 2023.
  7. W. Liang, G. A. Tadesse, D. Ho, L. Fei-Fei, M. Zaharia, C. Zhang, and J. Zou, “Advances, challenges and opportunities in creating data for trustworthy ai,” Nature Machine Intelligence, vol. 4, no. 8, pp. 669–677, 2022.
  8. O. Parraga, M. D. More, C. M. Oliveira, N. S. Gavenski, L. S. Kupssinskü, A. Medronha, L. V. Moura, G. S. Simões, and R. C. Barros, “Fairness in deep learning: A survey on vision and language research,” ACM Computing Surveys, 2023.
  9. S. Kolisko and C. J. Anderson, “Exploring social biases of large language models in a college artificial intelligence course,” Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 13, pp. 15 825–15 833, Sep. 2023. [Online]. Available: https://ojs.aaai.org/index.php/AAAI/article/view/26879
  10. K. Oksuz, B. C. Cam, S. Kalkan, and E. Akbas, “Imbalance problems in object detection: A review,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 10, pp. 3388–3415, 2021.
  11. Y.-Y. He, J. Wu, and X.-S. Wei, “Distilling virtual examples for long-tailed recognition,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 235–244.
  12. Y. Hong, S. Han, K. Choi, S. Seo, B. Kim, and B. Chang, “Disentangling label distribution for long-tailed visual recognition,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 6626–6636.
  13. J. Ren, C. Yu, X. Ma, H. Zhao, S. Yi et al., “Balanced meta-softmax for long-tailed visual recognition,” Advances in neural information processing systems, vol. 33, pp. 4175–4186, 2020.
  14. Y. Ma, L. Jiao, F. Liu, S. Yang, X. Liu, and P. Chen, “Feature distribution representation learning based on knowledge transfer for long-tailed classification,” IEEE Transactions on Multimedia, vol. 26, pp. 2772–2784, 2024.
  15. ——, “Geometric prior guided feature representation learning for long-tailed classification,” International Journal of Computer Vision, pp. 1–18, 2024.
  16. S. Sinha, H. Ohashi, and K. Nakamura, “Class-difficulty based methods for long-tailed visual recognition,” International Journal of Computer Vision, vol. 130, no. 10, pp. 2517–2531, 2022.
  17. Y. Ma, L. Jiao, F. Liu, Y. Li, S. Yang, and X. Liu, “Delving into semantic scale imbalance,” in The Eleventh International Conference on Learning Representations, 2023. [Online]. Available: https://openreview.net/forum?id=07tc5kKRIo
  18. Y. Ma, L. Jiao, F. Liu, S. Yang, X. Liu, and L. Li, “Curvature-balanced feature manifold learning for long-tailed classification,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 15 824–15 835.
  19. J. B. Tenenbaum, V. d. Silva, and J. C. Langford, “A global geometric framework for nonlinear dimensionality reduction,” science, vol. 290, no. 5500, pp. 2319–2323, 2000.
  20. N. Lei, D. An, Y. Guo, K. Su, S. Liu, Z. Luo, S.-T. Yau, and X. Gu, “A geometric understanding of deep learning,” Engineering, vol. 6, no. 3, pp. 361–374, 2020.
  21. Y. Ma, L. Jiao, F. Liu, S. Yang, X. Liu, and L. Li, “Orthogonal uncertainty representation of data manifold for robust long-tailed learning,” in Proceedings of the 31st ACM International Conference on Multimedia, 2023, pp. 4848–4857.
  22. U. Cohen, S. Chung, D. D. Lee, and H. Sompolinsky, “Separability and geometry of object manifolds in deep neural networks,” Nature communications, vol. 11, no. 1, p. 746, 2020.
  23. F. Camastra and A. Staiano, “Intrinsic dimension estimation: Advances and open problems,” Information Sciences, vol. 328, pp. 26–41, 2016.
  24. E. Levina and P. Bickel, “Maximum likelihood estimation of intrinsic dimension,” in Advances in Neural Information Processing Systems, L. Saul, Y. Weiss, and L. Bottou, Eds., vol. 17.   MIT Press, 2004. [Online]. Available: https://proceedings.neurips.cc/paper_files/paper/2004/file/74934548253bcab8490ebd74afed7031-Paper.pdf
  25. L. Amsaleg, O. Chelly, M. E. Houle, K.-i. Kawarabayashi, M. Radovanović, and W. Treeratanajaru, “Intrinsic dimensionality estimation within tight localities,” in Proceedings of the 2019 SIAM international conference on data mining.   SIAM, 2019, pp. 181–189.
  26. A. Litwin-Kumar, K. D. Harris, R. Axel, H. Sompolinsky, and L. Abbott, “Optimal degrees of synaptic connectivity,” Neuron, vol. 93, no. 5, pp. 1153–1164, 2017.
  27. K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” arXiv preprint arXiv:1409.1556, 2014.
  28. K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778.
  29. A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929, 2020.
  30. A. Krizhevsky, G. Hinton et al., “Learning multiple layers of features from tiny images,” 2009.
  31. O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein et al., “Imagenet large scale visual recognition challenge,” International journal of computer vision, vol. 115, pp. 211–252, 2015.
  32. L. Fei-Fei, R. Fergus, and P. Perona, “Learning generative visual models from few training examples: An incremental bayesian approach tested on 101 object categories,” in 2004 conference on computer vision and pattern recognition workshop.   IEEE, 2004, pp. 178–178.
  33. S. Chung, D. D. Lee, and H. Sompolinsky, “Classification and geometry of general perceptual manifolds,” Physical Review X, vol. 8, no. 3, p. 031003, 2018.
  34. M. S. Halvagal and F. Zenke, “The combination of hebbian and predictive plasticity learns invariant object representations in deep sensory networks,” Nature Neuroscience, vol. 26, no. 11, pp. 1906–1915, 2023.
  35. X. Li and S. Wang, “Toward a computational theory of manifold untangling: from global embedding to local flattening,” Frontiers in Computational Neuroscience, vol. 17, p. 1197031, 2023.
  36. M. T. Islam, Z. Zhou, H. Ren, M. B. Khuzani, D. Kapp, J. Zou, L. Tian, J. C. Liao, and L. Xing, “Revealing hidden patterns in deep neural network feature space continuum via manifold learning,” Nature Communications, vol. 14, no. 1, p. 8506, 2023.
  37. A. Ansuini, A. Laio, J. H. Macke, and D. Zoccolan, “Intrinsic dimension of data representations in deep neural networks,” Advances in Neural Information Processing Systems, vol. 32, 2019.
  38. P. Pope, C. Zhu, A. Abdelkader, M. Goldblum, and T. Goldstein, “The intrinsic dimension of images and its impact on learning,” in International Conference on Learning Representations, 2021. [Online]. Available: https://openreview.net/forum?id=XJk19XzGq2J
  39. Y. Cui, M. Jia, T.-Y. Lin, Y. Song, and S. Belongie, “Class-balanced loss based on effective number of samples,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 9268–9277.
  40. B. Kang, S. Xie, M. Rohrbach, Z. Yan, A. Gordo, J. Feng, and Y. Kalantidis, “Decoupling representation and classifier for long-tailed recognition,” arXiv preprint arXiv:1910.09217, 2019.
  41. B. Zhou, Q. Cui, X.-S. Wei, and Z.-M. Chen, “Bbn: Bilateral-branch network with cumulative learning for long-tailed visual recognition,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 9719–9728.
  42. X. Wang, H. Zhang, W. Huang, and M. R. Scott, “Cross-batch memory for embedding learning,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 6388–6397.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (8)
  1. Yanbiao Ma (13 papers)
  2. Licheng Jiao (109 papers)
  3. Fang Liu (801 papers)
  4. Lingling Li (34 papers)
  5. Wenping Ma (25 papers)
  6. Shuyuan Yang (36 papers)
  7. Xu Liu (213 papers)
  8. Puhua Chen (10 papers)

Summary

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

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