A Secure Federated Learning Approach: Preventing Model Poisoning Attacks via Optimal Clustering
Bibliographic record
Abstract
Federated Learning (FL) is a machine learning architecture that enables mnay participants to train a single machine learning model while preserving the security and privacy of each participant. FL is vulnerable to model poisoning attacks where an attacker sends poisoned model updates to compromise the global model. Existing defenses such as Byzantine-robust methods or malicious detection systems attempt to defend the global model from attackers. However, they can only resist a small number of malicious clients and attacks. In this work, we present an analysis on the latest defense FLDetector and propose an improved method. One issue with this method is that FLDetector always clusters clients into two clusters when it can regardless of the optimal cluster count. This causes it to misclassify clients to the wrong label resulting in removing the wrong clients from the training phase. This prevents the machine learning model from learning valuable information while allowing malicious clients to continue attacking the global model. The proposed approach utilizes Gap statistics to identify the ideal number of groups to separate the users. This enhances the filtering systems for attackers and reduces the false labelling of legitimate users in a Federated Learning environment. Our experimental results demonstrate an improvement compared to the baseline approach.
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 | 0.004 | 0.010 |
| 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 | 0.002 |
| Scholarly communication | 0.002 | 0.004 |
| Open science | 0.003 | 0.005 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 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".