A Privacy-Aware and Incremental Defense Method Against GAN-Based Poisoning Attack
The DOI matches its Crossref record; 42 of the 42 references it cites were found in a registry or an index.
Abstract
Federated learning is usually utilized as a fraud detection framework in the domain of financial risk management, which promotes the model accuracy without training data exchange. One of the challenges in federated learning is the GAN-based poisoning attack. The GAN-based poisoning attack is a type of intractable poisoning attack that causes global model accuracy degradation and privacy leak. Most of the existing defenses for GAN-based poisoning attack have the three problems: 1) dependence on validation datasets; 2) incompetence of dealing with incremental poisoning attack; and 3) privacy leak. To address the above problems, we present a privacy-aware and incremental defense (PID) method to detect malicious participants and protect privacy. In PID, we design a method to accumulate the offset of model parameters from participants in all current epochs to represent the moving tendency for model parameters. Thus, we can distinguish the adversaries from normal participants based on the accumulations in this incremental poisoning attack. We also use multiple trust domains to reduce the rate of misjudging benign participants as adversaries. Moreover, a differentiated differential privacy is utilized before the global model sending to protect the privacy of participants’ training datasets in PID. The experiments conducted on two real-world datasets under financial fraud detection scenario demonstrate that the PID reduces the fallout of adversaries detection (the rate of misjudging benign participants as adversaries) by at least 51.1% and improve the speed of detecting all malicious participants by at least 33.4% compared with two popular defense methods. Besides, the privacy preserving of PID is also effective.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Algorithm AS 136: A K-Means Clustering Algorithm
Hartigan, J. A., & Wong, M. A. (1979) · Applied Statistics · doi:10.2307/2346830 · Cited by 14,523
Clustering by Passing Messages Between Data Points
Frey, B. J., & Dueck, D. (2007) · Science · doi:10.1126/science.1136800 · Cited by 6,758 · cited 2×
“The idea of AP algorithm [37] is selecting cluster centers by passing information between different points.”
BackgroundDeep Learning with Differential Privacy
Abadi, M., Chu, A., Goodfellow, I., McMahan, H. B., Mironov, I., Talwar, K., & Zhang, L. (2016) · Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security · doi:10.1145/2976749.2978318 · Cited by 6,578 · cited 1×
“Definition 3 ( (ϵ, δ) -Differential Privacy [38]): With the domain X and range R , a randomized mechanism M : X → R satisfying (ϵ, δ) -differential privacy means that any two different inputs x , x ′ ∈ X and any subsets of outputs S ⊆ R have where where the j th sample in dataset D i is denoted by…”
BackgroundFederated learning: Challenges, methods, and future directions
Li, T., Sahu, A. K., Talwalkar, A., & Smith, V. (2020) · IEEE Signal Processing Magazine · doi:10.1109/msp.2020.2975749 · Cited by 6,210 · cited 1×
“Though federated learning has made extraordinary progress in training financial risk control model, there are still a series of concerns [8].”
BackgroundThe Algorithmic Foundations of Differential Privacy
Dwork, C., & Roth, A. (2013) · doi:10.1561/9781601988195 · Cited by 3,856
DBSCAN Revisited, Revisited: Why and How You Should (Still) Use DBSCAN
Schubert, E., Sander, J., Ester, M., Kriegel, H. P., & Xu, X. (2017) · ACM Transactions on Database Systems · doi:10.1145/3068335 · Cited by 2,737
How to backdoor federated learning
Bagdasaryan, E., Veit, A., Hua, Y., Estrin, D., & Shmatikov, V. (2018) · arXiv · doi:10.48550/arxiv.1807.00459 · Cited by 1,495 · cited 1×
“While data poisoning attacks [24] are easier to launch but difficult to defense.”
BackgroundByzantine-Robust Distributed Learning: Towards Optimal Statistical Rates
Yin, D., Chen, Y., Ramchandran, K., & Bartlett, P. (2018) · arXiv · doi:10.48550/arxiv.1803.01498 · Cited by 1,130 · cited 1×
“Yin et al. [26] designed a trimmed-mean-based model aggregation algorithm.”
MethodLocal Model Poisoning Attacks to Byzantine-Robust Federated Learning
Fang, M., Cao, X., Jia, J., & Gong, N. Z. (2019) · arXiv · doi:10.48550/arxiv.1911.11815 · Cited by 828 · cited 1×
“In federated learning, model poisoning attackers poison model parameters directly to launch attacks [23].”
BackgroundThe Fréchet distance between multivariate normal distributions
Dowson, D. C., & Landau, B. V. (1982) · Journal of Multivariate Analysis · doi:10.1016/0047-259x(82)90077-x · Cited by 674
Reliable Federated Learning for Mobile Networks
Kang, J., Xiong, Z., Niyato, D., Zou, Y., Zhang, Y., & Guizani, M. (2020) · IEEE Wireless Communications · doi:10.1109/mwc.001.1900119 · Cited by 617 · cited 1×
“In literature, Kang et al. [30] describe the concept of reputation in federated learning, which can judge whether the model parameters are reliable.”
BackgroundAttack of the tails: Yes, you really can backdoor federated learning
Wang, H., Sreenivasan, K., Rajput, S., Vishwakarma, H., Agarwal, S., Sohn, J.-Y., Lee, K., & Papailiopoulos, D. (2020) · arXiv · doi:10.48550/arxiv.2007.05084 · Cited by 406
Distributed Statistical Machine Learning in Adversarial Settings: Byzantine Gradient Descent
Chen, Y., Su, L., & Xu, J. (2017) · Proceedings of the ACM on Measurement and Analysis of Computing Systems · doi:10.1145/3154503 · Cited by 390
Privacy-Enhanced Federated Learning Against Poisoning Adversaries
Liu, X., Li, H., Xu, G., Chen, Z., Huang, X., & Lu, R. (2021) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2021.3108434 · Cited by 351 · cited 1×
“Liu et al. [28] utilize homomorphic encryption and extract the gradient of the logarithmic function to detect and penalize poisoners.”
BackgroundPoisonGAN: Generative Poisoning Attacks Against Federated Learning in Edge Computing Systems
Zhang, J., Chen, B., Cheng, X., Binh, H. T. T., & Yu, S. (2021) · IEEE Internet of Things Journal · doi:10.1109/jiot.2020.3023126 · Cited by 291 · cited 2×
“GAN-based poisoning attack [9], [10] is a burgeoning and intractable poisoning attack at present.”
BackgroundFederated Learning for Open Banking
Long, G., Tan, Y., Jiang, J., & Zhang, C. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-63076-8_17 · Cited by 244 · cited 1×
“Therefore, lots of financial fraud detection systems use federated learning framework currently [6], [7].”
BackgroundByzantine Attack and Defense in Cognitive Radio Networks: A Survey
Zhang, L., Ding, G., Wu, Q., Zou, Y., Han, Z., & Wang, J. (2015) · IEEE Communications Surveys & Tutorials · doi:10.1109/comst.2015.2422735 · Cited by 222 · cited 1×
“There are some researches on defense of some classic attacks, such as the backdoor attack [11], [12], the label-flipping attack [13], and the Byzantine attack [14].”
BackgroundFed-IIoT: A Robust Federated Malware Detection Architecture in Industrial IoT
Taheri, R., Shojafar, M., Alazab, M., & Tafazolli, R. (2021) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2020.3043458 · Cited by 218 · cited 6×
“Moreover, we also added the experiments compared with the algorithms in [15] and [16].”
BackgroundMethodInfluentialPoisoning Attack in Federated Learning using Generative Adversarial Nets
Zhang, J., Chen, J., Wu, D., Chen, B., & Yu, S. (2019) · 2019 18th IEEE International Conference On Trust, Security And Privacy In Computing And Communications/13th IEEE International Conference On Big Data Science And Engineering (TrustCom/BigDataSE) · doi:10.1109/trustcom/bigdatase.2019.00057 · Cited by 197 · cited 2×
“GAN-based poisoning attack [9], [10] is a burgeoning and intractable poisoning attack at present.”
BackgroundFederated learning for privacy-preserving AI
Cheng, Y., Liu, Y., Chen, T., & Yang, Q. (2020) · Communications of the ACM · doi:10.1145/3387107 · Cited by 186 · cited 1×
“Therefore, lots of financial fraud detection systems use federated learning framework currently [6], [7].”
BackgroundCredit Card Fraud Detection: A Novel Approach Using Aggregation Strategy and Feedback Mechanism
Jiang, C., Song, J., Liu, G., Zheng, L., & Luan, W. (2018) · IEEE Internet of Things Journal · doi:10.1109/jiot.2018.2816007 · Cited by 183 · cited 1×
“2) System Settings: Under the financial antifraud scenario [42], [43], we assume that there are 30 participants and a center server.”
BackgroundLi, Z., Huang, M., Liu, G., & Jiang, C. (2021) · Expert Systems with Applications · doi:10.1016/j.eswa.2021.114750 · Cited by 175 · cited 1×
“The fraud detection data are strongly imbalanced with most of the data being normal [44], [45].”
BackgroundSelf-citationLoMar: A Local Defense Against Poisoning Attack on Federated Learning
Li, X., Qu, Z., Zhao, S., Tang, B., Lu, Z., & Liu, Y. (2023) · IEEE Transactions on Dependable and Secure Computing · doi:10.1109/tdsc.2021.3135422 · Cited by 149 · cited 1×
“Li et al. in literature [31] presented a two-phase defense algorithm.”
BackgroundDeep Representation Learning With Full Center Loss for Credit Card Fraud Detection
Li, Z., Liu, G., & Jiang, C. (2020) · IEEE Transactions on Computational Social Systems · doi:10.1109/tcss.2020.2970805 · Cited by 144 · cited 1×
“Digital Object Identifier 10.1109/TCSS.2023.3263241 control [1], [2], [3], [4], [5].”
InfluentialSelf-citationTransaction Fraud Detection Based on Total Order Relation and Behavior Diversity
Zheng, L., Liu, G., Yan, C., & Jiang, C. (2018) · IEEE Transactions on Computational Social Systems · doi:10.1109/tcss.2018.2856910 · Cited by 138 · cited 1×
“Digital Object Identifier 10.1109/TCSS.2023.3263241 control [1], [2], [3], [4], [5].”
InfluentialXie, Y., Liu, G., Yan, C., Jiang, C., & Zhou, M. (2022) · IEEE Transactions on Computational Social Systems · doi:10.1109/tcss.2022.3158318 · Cited by 135 · cited 1×
“Digital Object Identifier 10.1109/TCSS.2023.3263241 control [1], [2], [3], [4], [5].”
InfluentialLearning Transactional Behavioral Representations for Credit Card Fraud Detection
Xie, Y., Liu, G., Yan, C., Jiang, C., Zhou, M., & Li, M. (2022) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2022.3208967 · Cited by 135 · cited 1×
“Digital Object Identifier 10.1109/TCSS.2023.3263241 control [1], [2], [3], [4], [5].”
InfluentialGAN Enhanced Membership Inference: A Passive Local Attack in Federated Learning
Zhang, J., Zhang, J., Chen, J., & Yu, S. (2020) · ICC 2020 - 2020 IEEE International Conference on Communications (ICC) · doi:10.1109/icc40277.2020.9148790 · Cited by 133 · cited 1×
“Thus, adversaries can utilize GAN to generate fake data close to the distribution of other participants’ data, which can become training datasets for other attacks of privacy leak, such as the membership inference attack [18], [19].”
BackgroundA Differentially Private Federated Learning Model against Poisoning Attacks in Edge Computing
Zhou, J., Wu, N., Wang, Y., Gu, S., Cao, Z., Dong, X., & Choo, K. R. (2022) · IEEE Transactions on Dependable and Secure Computing · doi:10.1109/tdsc.2022.3168556 · Cited by 124 · cited 3×
“In literature, Zhou et al. [33] proposed a novel differential privacy algorithm against poisoning attacks in federated learning.”
BackgroundInfluentialImproved TrAdaBoost and its Application to Transaction Fraud Detection
Zheng, L., Liu, G., Yan, C., Jiang, C., Zhou, M., & Li, M. (2020) · IEEE Transactions on Computational Social Systems · doi:10.1109/tcss.2020.3017013 · Cited by 119 · cited 1×
“Digital Object Identifier 10.1109/TCSS.2023.3263241 control [1], [2], [3], [4], [5].”
InfluentialDetecting Malicious Model Updates from Federated Learning on Conditional Variational Autoencoder
Gu, Z., & Yang, Y. (2021) · 2021 IEEE International Parallel and Distributed Processing Symposium (IPDPS) · doi:10.1109/ipdps49936.2021.00075 · Cited by 47 · cited 1×
“In study, Gu and Yang [32] designed an anomaly detection framework based on conditional variational autoencoder.”
MethodToward Smart Security Enhancement of Federated Learning Networks
Tan, J., Liang, Y.-C., Luong, N. C., & Niyato, D. (2021) · IEEE Network · doi:10.1109/mnet.011.2000379 · Cited by 42 · cited 1×
“In literature, Tan et al. [29] presented a verify-before-aggregate algorithm.”
BackgroundExploiting Defenses against GAN-Based Feature Inference Attacks in Federated Learning
Luo, X., & Zhang, X. (2025) · ACM Transactions on Knowledge Discovery from Data · doi:10.1145/3719350 · Cited by 40 · cited 1×
“Though the studies in literature [20], [21], [22] discuss the privacy leak caused by GAN and prevent GAN from generating fake data close to the true data distribution by adding noise at the participant end.”
BackgroundYang, C., Liu, G., Yan, C., & Jiang, C. (2021) · Science China Information Sciences · doi:10.1007/s11432-019-2739-2 · Cited by 34 · cited 1×
“There are multifarious clustering algorithms, such as k-means algorithm (KMA) [34], [35], density-based spatial clustering of applications with noise algorithm (DBSCAN) [36], and affinity propagation clustering algorithm.”
MethodDefending against Backdoors in Federated Learning with Robust Learning Rate
Ozdayi, M. S., Kantarcioglu, M., & Gel, Y. R. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i10.17118 · Cited by 30
MANE: Model-Agnostic Non-linear Explanations for Deep Learning Model
Tian, Y. X., & Liu, G. (2020) · IEEE World Congress on Services (SERVICES) · doi:10.1109/services48979.2020.00021 · Cited by 30 · cited 1×
“2) System Settings: Under the financial antifraud scenario [42], [43], we assume that there are 30 participants and a center server.”
BackgroundVy, N. C., Quyen, N. H., Duy, P. T., & Pham, V.-H. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-92708-0_8 · Cited by 25 · cited 5×
“Vy et al. [13] presented a validation function that is applied before model aggregation.”
BackgroundMethodInfluentialGANMIA: GAN-based Black-box Membership Inference Attack
Bai, Y., Chen, D., Chen, T., & Fan, M. (2021) · International Conference on Communications · doi:10.1109/icc42927.2021.9500657 · Cited by 19 · cited 1×
“Thus, adversaries can utilize GAN to generate fake data close to the distribution of other participants’ data, which can become training datasets for other attacks of privacy leak, such as the membership inference attack [18], [19].”
BackgroundGAN-Based Information Leakage Attack Detection in Federated Learning
Lai, J., Huang, X., Gao, X., Xia, C., & Hua, J. (2022) · Security and Communication Networks · doi:10.1155/2022/4835776 · Cited by 16 · cited 2×
“The machine-learning models mentioned above regarding the proposed algorithm, [13], [15], [16] and [17] are the same and we show the model structures in Table I.”
BackgroundMethodModel Partition Defense against GAN Attacks on Collaborative Learning via Mobile Edge Computing
Ching, C.-W., Lin, T.-C., Chang, K.-H., Yao, C.-C., & Kuo, J.-J. (2020) · Globecom · doi:10.1109/globecom42002.2020.9322591 · Cited by 15 · cited 5×
“Moreover, we also added the experiments compared with the algorithms in [15] and [16].”
BackgroundMethodInfluentialDefending against Membership Inference Attacks in Federated learning via Adversarial Example
Xie, Y., Chen, B., Zhang, J., & Wu, D. (2021) · International Conference on Mobility, Sensing and Networking (MSN) · doi:10.1109/msn53354.2021.00036 · Cited by 11 · cited 1×
“Though the studies in literature [20], [21], [22] discuss the privacy leak caused by GAN and prevent GAN from generating fake data close to the true data distribution by adding noise at the participant end.”
BackgroundLiu, Z., Li, R., Miao, D., Ren, L., & Zhao, Y. (2022) · Security and Communication Networks · doi:10.1155/2022/1615476 · Cited by 5 · cited 1×
“Though the studies in literature [20], [21], [22] discuss the privacy leak caused by GAN and prevent GAN from generating fake data close to the true data distribution by adding noise at the participant end.”
Background
Research works printed but not linked: 5
- Blanchard, Machine learning with adversaries: Byzantine tolerant gradient descent, 30, 1
- 2022, Credit Card Fraud Detection
- 2022, Ccfraud
- 10.1016/j.neucom.2020.04.078 · doi:10.1016/j.neucom.2020.04.078
- Balle, Improving the Gaussian mechanism for differential privacy: Analytical calibration and optimal denoising, 394
Cited by
(1996) · Institutional Research Information System University of Turin (University of Turin) · doi:10.1111/j · Cited by 5,486
A Novel Blockchain-Based Deepfake Detection Method Using Federated and Deep Learning Models
Heidari, A., Navimipour, N. J., Dag, H., Talebi, S., & Unal, M. (2024) · Cognitive Computation · doi:10.1007/s12559-024-10255-7 · Cited by 164
Metadata
- Authors
- Qiao, F., Li, Z., & Kong, Y.
- Venue
- IEEE Transactions on Computational Social Systems
- Published
- 2024-04
- Volume
- 11
- Issue
- 2
- Pages
- 1708-1721
- ISSN
- 2329-924X (ISSN-L)
- 2373-7476
- OpenAlex
- W4365790357
- Cited by
- 221
- References
- 42
- FWCI
- 3.07
- Type
- journal article
- Language
- en
- Open access
- closed
- Licence
- Funders
- National Natural Science Foundation of China (61972080)
- Natural Science Foundation of Shanghai (23ZR1401000)
- Shanghai Science and Technology Innovation Action Plan Project (22511100700)
- Open Foundation of the Key Laboratory of the Ministry of Education for Embedded System and Service Computing, Tongji University (ESSCKF2021-01)
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.