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

Debiasing Machine Unlearning with Counterfactual Examples (2404.15760v1)

Published 24 Apr 2024 in cs.LG, cs.AI, and stat.ML

Abstract: The right to be forgotten (RTBF) seeks to safeguard individuals from the enduring effects of their historical actions by implementing machine-learning techniques. These techniques facilitate the deletion of previously acquired knowledge without requiring extensive model retraining. However, they often overlook a critical issue: unlearning processes bias. This bias emerges from two main sources: (1) data-level bias, characterized by uneven data removal, and (2) algorithm-level bias, which leads to the contamination of the remaining dataset, thereby degrading model accuracy. In this work, we analyze the causal factors behind the unlearning process and mitigate biases at both data and algorithmic levels. Typically, we introduce an intervention-based approach, where knowledge to forget is erased with a debiased dataset. Besides, we guide the forgetting procedure by leveraging counterfactual examples, as they maintain semantic data consistency without hurting performance on the remaining dataset. Experimental results demonstrate that our method outperforms existing machine unlearning baselines on evaluation metrics.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (62)
  1. Meaningfully debugging model mistakes using conceptual counterfactual explanations. In International Conference on Machine Learning, pages 66–88. PMLR, 2022.
  2. What’s up with requirements engineering for artificial intelligence systems? In 2021 IEEE 29th International Requirements Engineering Conference (RE), pages 1–12. IEEE, 2021.
  3. Machine unlearning. In 2021 IEEE Symposium on Security and Privacy (SP), pages 141–159. IEEE, 2021.
  4. Towards making systems forget with machine unlearning. In 2015 IEEE symposium on security and privacy, pages 463–480. IEEE, 2015.
  5. Multi-agent covering option discovery based on kronecker product of factor graphs. IEEE Transactions on Artificial Intelligence, 2022.
  6. Explainable learning-based intrusion detection supported by memristors. In 2023 IEEE Conference on Artificial Intelligence (CAI), pages 195–196. IEEE, 2023.
  7. Boundary unlearning: Rapid forgetting of deep networks via shifting the decision boundary. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 7766–7775, 2023.
  8. A novel online incremental and decremental learning algorithm based on variable support vector machine. Cluster Computing, 22:7435–7445, 2019.
  9. Task-driven causal feature distillation: Towards trustworthy risk prediction. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 11642–11650, 2024.
  10. Can bad teaching induce forgetting? unlearning in deep networks using an incompetent teacher. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, pages 7210–7217, 2023.
  11. Variational nested dropout. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023.
  12. On the separability of classes with the cross-entropy loss function. arXiv preprint arXiv:1909.06930, 2019.
  13. A survey on bias in visual datasets. Computer Vision and Image Understanding, 223:103552, 2022.
  14. Dlora: Distributed parameter-efficient fine-tuning solution for large language model. arXiv preprint arXiv:2404.05182, 2024.
  15. Shortcut learning in deep neural networks. Nature Machine Intelligence, 2(11):665–673, 2020.
  16. Ufid: A unified framework for input-level backdoor detection on diffusion models. arXiv preprint arXiv:2404.01101, 2024.
  17. Local rule-based explanations of black box decision systems. arXiv preprint arXiv:1805.10820, 2018.
  18. Equality of opportunity in supervised learning. Advances in neural information processing systems, 29, 2016.
  19. A hierarchical spatial transformer for massive point samples in continuous space. Advances in Neural Information Processing Systems, 36, 2024.
  20. Unlearnable examples: Making personal data unexploitable. arXiv preprint arXiv:2101.04898, 2021.
  21. Fairsisa: Ensemble post-processing to improve fairness of unlearning in llms. arXiv preprint arXiv:2312.07420, 2023.
  22. Bridging adversarial robustness and gradient interpretability. arXiv preprint arXiv:1903.11626, 2019.
  23. Towards unbounded machine unlearning. Advances in Neural Information Processing Systems, 36, 2024.
  24. Towards unbounded machine unlearning. arXiv preprint arXiv:2302.09880, 2023.
  25. A combinatorial algorithm for approximating the optimal transport in the parallel and mpc settings. Advances in Neural Information Processing Systems, 36, 2024.
  26. Revisiting frequency analysis against encrypted deduplication via statistical distribution. In IEEE INFOCOM 2022-IEEE Conference on Computer Communications, pages 290–299. IEEE, 2022.
  27. Please tell me more: Privacy impact of explainability through the lens of membership inference attack. In 2024 IEEE Symposium on Security and Privacy (SP), pages 120–120. IEEE Computer Society, 2024.
  28. Riatig: Reliable and imperceptible adversarial text-to-image generation with natural prompts. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20585–20594, 2023.
  29. Neural-answering logical queries on knowledge graphs. In Proceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining, pages 1087–1097, 2021.
  30. Focus: Flexible optimizable counterfactual explanations for tree ensembles. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 5313–5322, 2022.
  31. A study of the attention abnormality in trojaned berts. arXiv preprint arXiv:2205.08305, 2022.
  32. Attention-enhancing backdoor attacks against bert-based models. arXiv preprint arXiv:2310.14480, 2023.
  33. Learning individualized treatment rules with many treatments: A supervised clustering approach using adaptive fusion. Advances in Neural Information Processing Systems, 35:15956–15969, 2022.
  34. Learning optimal group-structured individualized treatment rules with many treatments. Journal of Machine Learning Research, 24(102):1–48, 2023.
  35. Data-driven transfer learning framework for estimating on-ramp and off-ramp traffic flows. Journal of Intelligent Transportation Systems, pages 1–14, 2024.
  36. Less is more: Understanding network bias in proof-of-work blockchains. Mathematics, 11(23):4741, 2023.
  37. Accelerating general-purpose lossless compression via simple and scalable parameterization. In Proceedings of the 30th ACM International Conference on Multimedia, pages 3205–3213, 2022.
  38. Trace: A fast transformer-based general-purpose lossless compressor. In Proceedings of the ACM Web Conference 2022, pages 1829–1838, 2022.
  39. A survey of machine unlearning. arXiv preprint arXiv:2209.02299, 2022.
  40. Fair machine unlearning: Data removal while mitigating disparities. arXiv preprint arXiv:2307.14754, 2023.
  41. Learning model-agnostic counterfactual explanations for tabular data. In Proceedings of The Web Conference 2020, pages 3126–3132, 2020.
  42. Judea Pearl. Causal inference in statistics: An overview. Statistics Surveys, 3:96–146, 2009.
  43. Ssse: Efficiently erasing samples from trained machine learning models. arXiv preprint arXiv:2107.03860, 2021.
  44. Counterfactual interpolation augmentation (cia): A unified approach to enhance fairness and explainability of dnn. In IJCAI, pages 732–739, 2022.
  45. Dice: Domain-attack invariant causal learning for improved data privacy protection and adversarial robustness. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 1483–1492, 2022.
  46. Learning human driving behaviors with sequential causal imitation learning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 4583–4592, 2022.
  47. Causal imitation learning via inverse reinforcement learning. In The Eleventh International Conference on Learning Representations, 2022.
  48. A Survey of Contrastive and Counterfactual Explanation Generation Methods for Explainable Artificial Intelligence. IEEE Access, 9:11974–12001, 2021.
  49. Contrastive boundary learning for point cloud segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8489–8499, 2022.
  50. Debiasing nlu models via causal intervention and counterfactual reasoning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 11376–11384, 2022.
  51. Automated directed fairness testing. In Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering, pages 98–108, 2018.
  52. Making heads or tails: Towards semantically consistent visual counterfactuals. In European Conference on Computer Vision, pages 261–279. Springer, 2022.
  53. Puma: Performance unchanged model augmentation for training data removal. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 8675–8682, 2022.
  54. Tkil: Tangent kernel optimization for class balanced incremental learning. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 3529–3539, 2023.
  55. A measurement study of fmcw radar configurations for non-contact vital signs monitoring. In 2022 IEEE Radar Conference (RadarConf22), pages 1–6. IEEE, 2022.
  56. Vitalhub: Robust, non-touch multi-user vital signs monitoring using depth camera-aided uwb. In 2021 IEEE 9th International Conference on Healthcare Informatics (ICHI), pages 320–329. IEEE, 2021.
  57. Medlens: Improve mortality prediction via medical signs selecting and regression. In 2023 IEEE 3rd International Conference on Computer Communication and Artificial Intelligence (CCAI), pages 169–175. IEEE, 2023.
  58. Incremental learning meets transfer learning: Application to multi-site prostate mri segmentation. In International Workshop on Distributed, Collaborative, and Federated Learning, pages 3–16. Springer, 2022.
  59. Unleashing the power of self-supervised image denoising: A comprehensive review. arXiv preprint arXiv:2308.00247, 2023.
  60. To be forgotten or to be fair: Unveiling fairness implications of machine unlearning methods. AI and Ethics, pages 1–11, 2024.
  61. Stable and safe reinforcement learning via a barrier-lyapunov actor-critic approach. In 2023 62nd IEEE Conference on Decision and Control (CDC), pages 1320–1325. IEEE, 2023.
  62. Xai meets biology: A comprehensive review of explainable ai in bioinformatics applications. arXiv preprint arXiv:2312.06082, 2023.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (10)
  1. Ziheng Chen (30 papers)
  2. Jia Wang (163 papers)
  3. Jun Zhuang (34 papers)
  4. Abbavaram Gowtham Reddy (14 papers)
  5. Fabrizio Silvestri (75 papers)
  6. Jin Huang (80 papers)
  7. Kaushiki Nag (11 papers)
  8. Kun Kuang (114 papers)
  9. Xin Ning (22 papers)
  10. Gabriele Tolomei (26 papers)
Citations (2)
View on Semantic Scholar

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