An Explainable and Privacy-Preserving Federated Learning Model for Threat Detection in Cyber-Physical-Social Systems
Bibliographic record
Abstract
Cyber-physical-social systems (CPSS), the technological backbone of infrastructures such as energy, aerospace, transportation, and human-centric applications, are increasingly prone to cyberattacks due to their interconnected nature. Existing threat detection models focus on performance metrics, while effective, often overlook the reliability and privacy of data in CPSS and lack explainability in detecting anomalies. This article introduces an innovative federated learning (FL) model in threat detection for the resilience assessment and enhancement of CPSS. It is not only explainable and privacy-preserving (PP) but also ensures the reliable operation of systems by addressing their inherent vulnerabilities. It addresses current shortcomings, emphasizing the often-neglected accuracy and reliability in literature. In this regard, we reinforce privacy by utilizing differential privacy (DP) and partially homomorphic encryption (PHE), especially the Paillier homomorphic cryptosystem, for secure model update aggregation in the FL environment. The system’s resilience against malicious attempts is heightened by ensuring a reliable aggregation process. To enhance model transparency and facilitate effective responses to threats, we integrate SHapley Additive exPlanations (SHAP) for detailed predictive contribution breakdowns and discern relationships between CPSS’ sensors, cyber components, and social interactions for aberration detection. This improves system reliability by clarifying the contribution of each sensor, device, or social actor, and identifying potential points of failure. The efficacy of our model was rigorously evaluated on three representative CPSS datasets: Secure Water Treatment (SWaT), Gas Pipeline, and UNSW-NB15. It achieved anomaly detection accuracies of 97.46%, 98.85%, and 93.60%, respectively, outperforming several baseline methods. Our model maintains strong privacy guarantees through cryptographic techniques and provides meaningful explainability during anomaly detection, fostering greater trust and reliability in critical CPSS operations.
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 | 0.013 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.002 | 0.004 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
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".