RBFL: Securing Federated Learning against Data Poisoning Using a Reputation-based Approach
Bibliographic record
Abstract
Federated learning enables distributed clients to train a global model while maintaining privacy and control over their data. Although collaboration can substantially improve the learning process, it also introduces vulnerabilities, as not all participants contribute beneficially. Clients may engage in detrimental activities, such as data poisoning, that compromise the integrity of the global model. Additionally, in a realistic scenario, the quality of the data possessed by these clients is highly heterogeneous, which influences the model training performance. Moreover, malicious clients (aka free riders) intend to obtain the global model without making a real contribution to the training process. Hence, a reliable and fair evaluation of the client contribution is essential to promote diverse client engagement, improve robustness, and address the free-rider problem. This paper proposes a reputation-aware contribution evaluation approach (RBFL) that provides adversarial robustness by tracking reputation over multiple training rounds to ensure that clients consistently contribute positively. We employ CosineGradient as the utility function and Truncated Monte Carlo (TMC) Shapley as the data valuation function. Empirical evaluation demonstrates the effectiveness of our approach in a fair evaluation of clients’ contributions and effective identification of adversarial clients while maintaining a model accuracy of $92 \%$ with adversarial robustness.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.019 | 0.049 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.002 | 0.003 |
| Scholarly communication | 0.003 | 0.006 |
| Open science | 0.005 | 0.008 |
| Research integrity | 0.003 | 0.003 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".