Data Overvaluation Attack and Truthful Data Valuation in Federated Learning
Not on the map. Why?
Abstract (arXiv)
In collaborative machine learning (CML), data valuation, i.e., evaluating the contribution of each client's data to the machine learning model, has become a critical task for incentivizing and selecting positive data contributions. However, existing studies often assume that clients engage in data valuation truthfully, overlooking the practical motivation for clients to exaggerate their contributions. To unlock this threat, this paper introduces the data overvaluation attack, enabling strategic clients to have their data significantly overvalued in federated learning, a widely adopted paradigm for decentralized CML. Furthermore, we propose a Bayesian truthful data valuation metric, named Truth-Shapley. Truth-Shapley is the unique metric that guarantees some promising axioms for data valuation while ensuring that clients' optimal strategy is to perform truthful data valuation under certain conditions. Our experiments demonstrate the vulnerability of existing data valuation metrics to the proposed attack and validate the robustness and effectiveness of Truth-Shapley.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 29 of its 48 references that are in the map. First posted on arXiv 1 Feb 2025.
Builds on
29 of 48 references.
References · 29
- 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,049
- 2016Communication-Efficient Learning of Deep Networks from Decentralized Datacited by 7,592
- 2017Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithmscited by 4,283
- 2017Understanding Black-box Predictions via Influence Functionscited by 1,714
- 2019Data Shapley: Equitable Valuation of Data for Machine Learningcited by 617
- 2020Estimating Training Data Influence by Tracing Gradient Descentcited by 565
- 2019FedMD: Heterogenous Federated Learning via Model Distillationcited by 454
- 2020SplitFed: When Federated Learning Meets Split Learningcited by 308
- 2019Towards Efficient Data Valuation Based on the Shapley Valuecited by 279
- 2018Distributed learning of deep neural network over multiple agentscited by 277
- 2021Beta Shapley: a Unified and Noise-reduced Data Valuation Framework for Machine Learningcited by 162
- 2019Efficient Task-Specific Data Valuation for Nearest Neighbor Algorithmscited by 147
17 more
- 2022Data Banzhaf: A Robust Data Valuation Framework for Machine Learningcited by 126
- 2018A Marketplace for Data: An Algorithmic Solutioncited by 75
- 2020Collaborative Machine Learning with Incentive-Aware Model Rewardscited by 71
- 2023LAVA: Data Valuation without Pre-Specified Learning Algorithmscited by 69
- 2022Measuring the Effect of Training Data on Deep Learning Predictions via Randomized Experimentscited by 50
- 2020Efficient computation and analysis of distributional Shapley valuescited by 38
- 2021Game of Gradients: Mitigating Irrelevant Clients in Federated Learningcited by 35
- 2023Data Valuation Without Training of a Modelcited by 35
- 2022CS-Shapley: Class-wise Shapley Values for Data Valuation in Classificationcited by 32
- 2020Truthful Data Acquisition via Peer Predictioncited by 14
- 2019Collaborative Machine Learning Markets with Data-Replication-Robust Paymentscited by 13
- 2022Secure Shapley Value for Cross-Silo Federated Learning (Technical Report)cited by 13
- 2024Helpful or Harmful Data? Fine-tuning-free Shapley Attribution for Explaining Language Model Predictionscited by 13
- 2021FL-Market: Trading Private Models in Federated Learningcited by 9
- 2022On Shapley Value in Data Assemblage Under Independent Utilitycited by 4
- 2023DeRDaVa: Deletion-Robust Data Valuation for Machine Learningcited by 3
- 2024Privacy-Enhanced Database Synthesis for Benchmark Publishing (Technical Report)cited by 2
Built on by
Not cited yet.