On Privacy-Preserved Machine Learning Using Secure Multi-Party Computing: Techniques and Trends
Source abstract
Abstract
The rapid adoption of machine learning in sensitive domains, such as healthcare, finance, and government services, has heightened the need for robust, privacy-preserving techniques. Traditional machine learning approaches lack built-in privacy mechanisms, exposing sensitive data to risks, which motivates the development of Privacy-Preserving Machine Learning (PPML) methods. Despite significant advances in PPML, a comprehensive and focused exploration of Secure Multi-Party Computing (SMPC) within this context remains underdeveloped. This review aims to bridge this knowledge gap by systematically analyzing the role of SMPC in PPML, offering a structured overview of current techniques, challenges, and future directions. Using a semi-systematic mapping study methodology, this paper surveys recent literature spanning SMPC protocols, PPML frameworks, implementation approaches, threat models, and performance metrics. Emphasis is placed on identifying trends, technical limitations, and comparative strengths of leading SMPC-based methods. Our findings reveal that while SMPC offers strong cryptographic guarantees for privacy, challenges such as computational overhead, communication costs, and scalability persist. The paper also discusses critical vulnerabilities, practical deployment issues, and variations in protocol efficiency across use cases.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Privacy-Preserving Technologies in Data · Cryptography and Data Security · Network Security and Intrusion Detection
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References (100+)
Show source-provided references
1. Zhang C, Li S. State-of-the-art approaches to enhancing privacy preservation of machine learning datasets: a survey. arXiv:2404.16847. 2025.10.36227/techrxiv.170956580.07799070/v1 ↗
2. Efficient and secure multi-party computation protocol supporting deep learning10.1186/s42400-024-00343-4 ↗
3. Privacy-preservation for gradient descent methods
4. Privacy-preserving classifier learning
5. Pathak · Advances in neural information processing systems · 2010
6. Multiparty computation from somewhat homomorphic encryption
7. Secureml: a system for scalable privacy-preserving machine learning
8. Chameleon: a hybrid secure computation framework for machine learning applications
9. SecureNN: 3-party secure computation for neural network training10.2478/popets-2019-0035 ↗
10. QUOTIENT: two-party secure neural network training and prediction
11. Cryptflow: secure tensorflow inference
12. Crypten: secure multi-party computation meets machine learning
13. Wagh S, Tople S, Benhamouda F, Kushilevitz E, Mittal P, Rabin T. Falcon: honest-majority maliciously secure framework for private deep learning. arXiv:2004.02229. 2020. doi:10.48550/arxiv.2004.02229.10.2478/popets-2021-0011 ↗
14. Cheetah: lean and fast secure {two-party} deep neural network inference
15. Dong Y, Lu WJ, Zheng Y, Wu H, Zhao D, Tan J, et al. Puma: secure inference of llama-7b in five minutes. arXiv:2307.12533. 2023.
16. {SecretFlow-SPU}: a performant and {User-Friendly} framework for {Privacy-Preserving} machine learning
17. Jiang W, Song X, Hong S, Zhang H, Liu W, Zhao B, et al. Spin: an efficient secure computation framework with GPU acceleration. arXiv:2402.02320. 2024.
18. Peng S, Liu T, Tao T, Zhao D, Sheng H, Zhu H. EVA-S3PC: efficient, verifiable, accurate secure matrix multiplication protocol assembly and its application in regression. arXiv:2411.03404. 2024.
19. A survey on collaborative deep learning and privacy-preserving
20. A survey: privacy preservation techniques in data mining
21. Privacy preservation techniques in big data analytics: a survey10.1186/s40537-018-0141-8 ↗
22. Machine learning and genomics: precision medicine versus patient privacy10.1098/rsta.2017.0350 ↗
23. Distributed deep learning networks among institutions for medical imaging10.1093/jamia/ocy017 ↗
24. Security enhancement in healthcare cloud using machine learning10.1016/j.procs.2018.01.136 ↗
25. Developing high performance secure multi-party computation protocols in healthcare: a case study of patient risk stratification
26. A distributed fair machine learning framework with private demographic data protection
27. Rapid network planning of temporary private 5G networks with unsupervised machine learning
28. Privacy-preserving online AutoML for domain-specific face detection
29. TAPAS: tricks to accelerate (encrypted) prediction as a service
30. PrivEdge: from local to distributed private training and prediction
31. PPAAS: privacy preservation as a service10.1016/j.comcom.2021.04.006 ↗
32. Efficient secure multi-party computation for multi-dimensional arithmetics and its applications10.3390/cryptography9030050 ↗
33. Differential privacy: a survey of results
34. Trusted execution environment: what it is, and what it is not
35. Wu Z, Hou J, Diao Y, He B. Federated transformer: multi-party vertical federated learning on practical fuzzily linked data. arXiv:2410.17986. 2024.
36. Konečnỳ J, McMahan HB, Ramage D, Richtárik P. Federated optimization: distributed machine learning for on-device intelligence. arXiv:1610.02527. 2016.
37. Konečnỳ J, McMahan HB, Yu FX, Richtárik P, Suresh AT, Bacon D. Federated learning: strategies for improving communication efficiency. arXiv:1610.05492. 2016.
38. Zhao B, Mopuri KR, Bilen H. IDLG: improved deep leakage from gradients. arXiv:2001.02610. 2020.
39. Exploiting unintended feature leakage in collaborative learning
40. Zhu · Advances in neural information processing systems · 2019
41. Bettini C, Civitarese G, Presotto R. Personalized semi-supervised federated learning for human activity recognition. arXiv:2104.08094. 2021.
42. Samarati P, Sweeney L. Protecting privacy when disclosing information: k-anonymity and its enforcement through generalization and suppression [Internet]. Silicon Valley, CA, USA: SRI International, Massachusetts Institute of Technology; 1998 [cited 2025 Aug 12]. Available from: https://dataprivacylab.org/dataprivacy/projects/kanonymity/paper3.pdf.
43. k-anonymity: a model for protecting privacy10.1142/S0218488502001648 ↗
44. l-diversity: privacy beyond k-anonymity10.1145/1217299.1217302 ↗
45. t-closeness: privacy beyond k-anonymity and l-diversity
46. M-invariance: towards privacy preserving re-publication of dynamic datasets
47. Calibrating noise to sensitivity in private data analysis
48. Deep learning with differential privacy
49. Differential privacy preservation for deep auto-encoders: an application of human behavior prediction
50. Adaptive laplace mechanism: differential privacy preservation in deep learning
51. Preserving differential privacy in convolutional deep belief networks10.1007/s10994-017-5656-2 ↗
52. Papernot N, Abadi M, Erlingsson U, Goodfellow I, Talwar K. Semi-supervised knowledge transfer for deep learning from private training data. arXiv:1610.05755. 2016.
53. Papernot N, Song S, Mironov I, Raghunathan A, Talwar K, Erlingsson Ú. Scalable private learning with pate. arXiv:1802.08908. 2018.
54. Xie L, Lin K, Wang S, Wang F, Zhou J. Differentially private generative adversarial network. arXiv:1802.06739. 2018.
55. Differentially private mixture of generative neural networks10.1109/TKDE.2018.2855136 ↗
56. Deep learning with gaussian differential privacy
57. MPC vs. SFE: unconditional and computational security
58. Practical covertly secure MPC for dishonest majority-or: breaking the SPDZ limits
59. The round complexity of secure protocols
60. Efficient and private scoring of decision trees, support vector machines and logistic regression models based on pre-computation10.1109/TDSC.2017.2679189 ↗
61. SPDZ2k: efficient MPC mod 2k for dishonest majority
62. MASCOT: faster malicious arithmetic secure computation with oblivious transfer
63. MP-SPDZ: a versatile framework for multi-party computation
64. Demmler D, Schneider T, Zohner M. ABY-A framework for efficient mixed-protocol secure two-party computation [Internet]. In: NDSS; 2015 [cited 2025 Aug 12]. Available from: https://encrypto.de/papers/DSZ15.pdf.10.14722/ndss.2015.23113 ↗
65. EPIC: efficient private image classification (or: learning from the masters)
66. ASTRA: high throughput 3pc over rings with application to secure prediction
67. MP2ML: a mixed-protocol machine learning framework for private inference
68. Securegbm: secure multi-party gradient boosting
69. Mlcapsule: guarded offline deployment of machine learning as a service
70. Lightning-fast and privacy-preserving outsourced computation in the cloud10.1186/s42400-020-00057-3 ↗
71. Mscryptonet: multi-scheme privacy-preserving deep learning in cloud computing10.1109/ACCESS.2019.2901219 ↗
72. Cryptonets: applying neural networks to encrypted data with high throughput and accuracy
73. Xie P, Bilenko M, Finley T, Gilad-Bachrach R, Lauter K, Naehrig M. Crypto-nets: neural networks over encrypted data. arXiv:1412.6181. 2014. doi:10.48550/arxiv.1412.6181.
74. Hesamifard E, Takabi H, Ghasemi M. Cryptodl: deep neural networks over encrypted data. arXiv:1711.05189. 2017. doi:10.48550/arxiv.1711.05189.
75. Chabanne H, De Wargny A, Milgram J, Morel C, Prouff E. Privacy-preserving classification on deep neural network. Cryptology ePrint Archive 2017 [Internet]. [cited 2025 Aug 12]. Available from: https://eprint.iacr.org/2017/035.
76. Privacy-preserving distributed deep learning via homomorphic re-encryption10.3390/electronics8040411 ↗
77. Multi-key privacy-preserving deep learning in cloud computing10.1016/j.future.2017.02.006 ↗
78. Gu Z, Huang H, Zhang J, Su D, Lamba A, Pendarakis D, et al. Securing input data of deep learning inference systems via partitioned enclave execution. arXiv:1807.00969. 2018.
79. Hunt T, Song C, Shokri R, Shmatikov V, Witchel E. Chiron: privacy-preserving machine learning as a service. arXiv:1803.05961. 2018.
80. Ryoan: a distributed sandbox for untrusted computation on secret data10.1145/3231594 ↗
81. Tramer F, Boneh D. Slalom: fast, verifiable and private execution of neural networks in trusted hardware. arXiv:1806.03287. 2018.
82. On data banks and privacy homomorphisms
83. A survey on homomorphic encryption schemes: theory and implementation10.1145/3214303 ↗
84. CHET: an optimizing compiler for fully-homomorphic neural-network inferencing
85. ML confidential: machine learning on encrypted data
86. Privacy preserving multi-party machine learning with homomorphic encryption
87. Fast homomorphic evaluation of deep discretized neural networks
88. Han K, Hong S, Cheon JH, Park D. Efficient logistic regression on large encrypted data. Cryptology ePrint Archive 2018 [Internet]. [cited 2025 Aug 12]. Available from: https://eprint.iacr.org/2018/662.
89. nGraph-HE: a graph compiler for deep learning on homomorphically encrypted data
90. Folkerts L, Gouert C, Tsoutsos NG. REDsec: running encrypted discretized neural networks in seconds. Cryptology ePrint Archive 2021 [Internet]. [cited 2025 Aug 12]. Available from: https://eprint.iacr.org/2021/1100.
91. {AutoFHE}: automated adaption of {CNNs} for efficient evaluation over {FHE}
92. Keyword spotting in the homomorphic encrypted domain using deep complex-valued CNN
93. Zimerman I, Baruch M, Drucker N, Ezov G, Soceanu O, Wolf L. Converting transformers to polynomial form for secure inference over homomorphic encryption. arXiv:2311.08610. 2023.
94. ProteiNN: privacy-preserving one-to-many Neural Network classifications
95. Privacy-preserving deep learning via additively homomorphic encryption10.1109/TIFS.2017.2787987 ↗
96. Homomorphic encryption for machine learning in medicine and bioinformatics10.1145/3394658 ↗
97. Privacy preservation for machine learning training and classification based on homomorphic encryption schemes10.1016/j.ins.2020.03.041 ↗
98. Applying deep neural networks over homomorphic encrypted medical data10.1155/2020/3910250 ↗
99. Lattigo: a multiparty homomorphic encryption library in go
100. Fully homomorphic encryption using ideal lattices