Ensuring User Privacy and Model Security via Machine Unlearning: A Review
Source abstract
Abstract
As an emerging discipline, machine learning has been widely used in artificial intelligence, education, meteorology and other fields. In the training of machine learning models, trainers need to use a large amount of practical data, which inevitably involves user privacy. Besides, by polluting the training data, a malicious adversary can poison the model, thus compromising model security. The data provider hopes that the model trainer can prove to them the confidentiality of the model. Trainer will be required to withdraw data when the trust collapses. In the meantime, trainers hope to forget the injected data to regain security when finding crafted poisoned data after the model training. Therefore, we focus on forgetting systems, the process of which we call machine unlearning, capable of forgetting specific data entirely and efficiently. In this paper, we present the first comprehensive survey of this realm. We summarize and categorize existing machine unlearning methods based on their characteristics and analyze the relation between machine unlearning and relevant fields (e.g., inference attacks and data poisoning attacks). Finally, we briefly conclude the existing research directions.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Anomaly Detection Techniques and Applications · Adversarial Robustness in Machine Learning · Network Security and Intrusion Detection
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References (54)
Show source-provided references
1. Y. Liu, R. Wen, X. He, A. Salem, Z. Zhang et al., “ML-Doctor: Holistic risk assessment of inference attacks against machine learning models,” arXiv preprint arXiv:2102.02551, 2021.
2. Membership inference attacks against machine learning models
3. Council of European Union. Council regulation (EU) no 2012/0011, 2014. [Online]. Available: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52012PC0011 (accessed on 25/01/2012).
4. Council of European Union. Council regulation (EU) no 2016/678, 2014. [Online]. Available: https://eur-lex.europa.eu/eli/reg/2016/679/oj (accessed on 27/04/2016).
5. When machine unlearning jeopardizes privacy
6. H. Huang, J. Mu, N. Z. Gong, Q. Li, B. Liu et al., “Data poisoning attacks to deep learning based recommender systems,” arXiv preprint arXiv:2101.02644, 2021.10.14722/ndss.2021.24525 ↗
7. Towards making systems forget with machine unlearning
8. Ginart · 33rd Conf. on Neural Information Processing Systems · 2019 · 3518
9. Ensuring data confidentiality via plausibly deniable encryption and secure deletion–a survey10.1186/s42400-018-0005-8 ↗
10. Machine unlearning
11. Eternal sunshine of the spotless net: Selective forgetting in deep networks
12. S. Shintre and J. Dhaliwal, “Verifying that the influence of a user data point has been removed from a machine learning classifier,” Justia Patents, Patent # 10,225,277, 2019.
13. Formalizing data deletion in the context of the right to be forgotten
14. Forgetting outside the box: Scrubbing deep networks of information accessible from input-output observations
15. On the necessity of auditable algorithmic definitions for machine unlearning
16. A swiss army infinitesimal jackknife
17. T. Baumhauer, P. Schöttle and M. Zeppelzauer, “Machine unlearning: Linear filtration for logit-based classifiers,” arXiv preprint arXiv:2002.02730, 2020.
18. Mixed-privacy forgetting in deep networks
19. Approximate data deletion from machine learning models: Algorithms and evaluations
20. Learn to forget: Machine unlearning via neuron masking
21. Sekhari · Thirty-fifth Conf. on Neural Information Processing Systems · 2021 · 34
22. J. Brophy and D. Lowd, “DART: Data addition and removal trees,” arXiv preprint arXiv:2009.05567, 2020.
23. S. Fu, F. He, Y. Xu and D. Tao, “Bayesian inference forgetting,” arXiv preprint arXiv:2101.06417, 2021.
24. Variational bayesian unlearning
25. C. Guo, T. Goldstein, A. Hannun and L. van der Maaten, “Certified data removal from machine learning models,” arXiv preprint arXiv:1911.03030, 2019.
26. D. M. Sommer, L. Song, S. Wagh and P. Mittal, “Towards probabilistic verification of machine unlearning,” arXiv preprint arXiv:2003.04247, 2020.
27. Have you forgotten? A method to assess if machine learning models have forgotten data
28. Auditing data provenance in text-generation models
29. Enhancing Chinese character representation with lattice-aligned attention
30. A survey on adversarial example10.32604/jihpp.2020.010462 ↗
31. Reflection backdoor: A natural backdoor attack on deep neural networks
32. Opportunistic backdoor attacks: Exploring human-imperceptible vulnerabilities on speech recognition systems
33. Gupta · Thirty-fifth Conf. on Neural Information Processing Systems · 2021 · 34 · 16319
34. Machine learning privacy aware anonymization using mapreduce based neural network10.32604/iasc.2022.020164 ↗
35. Automated learning of ECG streaming data through machine learning Internet of Things10.32604/iasc.2022.021426 ↗
36. BadNets: Evaluating backdooring attacks on deep neural networks10.1109/ACCESS.2019.2909068 ↗
37. The secret sharer: Evaluating and testing unintended memorization in neural networks
38. G. Liu, X. Ma, Y. Yang, C. Wang and J. Liu, “Federated unlearning,” arXiv preprint arXiv:2012.13891, 2020.
39. Delta-grad: Rapid retraining of machine learning models
40. Neel · Algorithmic Learning Theory · 2021 · 931
41. M. Chen, Z. Zhang, T. Wang, M. Backes, M. Humbert et al., “Graph unlearning,” arXiv preprint arXiv:2103.14991, 2021.10.1145/3548606.3559352 ↗
42. A. Warnecke, L. Pirch, C. Wressnegger and K. Rieck, “Machine unlearning of features and labels,” arXiv preprint arXiv:2108.11577, 2021.
43. A. Thudi, G. Deza, V. Chandrasekaran and N. Papernot, “Unrolling sgd: Understanding factors influencing machine unlearning,” arXiv preprint arXiv:2109.13398, 2021.10.1109/EuroSP53844.2022.00027 ↗
44. A. Mahadevan and M. Mathioudakis, “Certifiable machine unlearning for linear models,” arXiv preprint arXiv:2106.15093, 2021.
45. Amnesia—A selection of machine learning models that can forget user data very fast
46. M. Jegorova, C. Kaul, C. Mayor, A. Q. O’Neil, A. Weir et al., “Survey: Leakage and privacy at inference time,” arXiv preprint arXiv:2107.01614, 2021.10.1109/TPAMI.2022.3229593 ↗
47. Publicly verifiable data transfer and deletion scheme for cloud storage10.1177/1550147719878999 ↗
48. Secure data deletion in cloud storage: A survey10.1504/IJES.2020.105939 ↗
49. Gotta catch’em all: Using honeypots to catch adversarial attacks on neural networks
50. Remote wiping and secure deletion on mobile devices: A review10.1111/1556-4029.13203 ↗
51. Machine unlearning via algorithmic stability
52. Lifelong anomaly detection through unlearning
53. Deletion-robust submodular maximization: Data summarization with “the right to be forgotten”
54. Hard to forget: Poisoning attacks on certified machine unlearning