| name | quantum-machine-learning-uav-anomaly-detection |
| description | Leakage-free evaluation of quantum ML for UAV anomaly detection. Group-aware temporal protocol + three-mode feature audit + hybrid XGBoost-DRU classifier. |
| category | quantum-ml |
Quantum ML for Cyber-Physical Anomaly Detection in UAVs
arXiv: 2605.19233 (cs.CR, cs.LG, quant-ph)
Authors: Carlos A. Durán Paredes, Javier E. León Calderón, Nicolás Sánchez Perea, German Darío Díaz, Camilo Segura Quintero
Core Methodology
Rigorous, leakage-free evaluation of quantum machine learning for unmanned aerial vehicle (UAV) anomaly detection on the multi-sensor TLM:UAV benchmark.
Three Key Contributions
-
Group-Aware Temporal Protocol (B2):
- Partitions dataset into 10 contiguous TimeUS blocks
- Evaluates over 10 seeds
- Eliminates inflation from random stratified splits that mix neighboring samples
-
Three-Mode Feature Audit (Full/Loose/Strict):
- Quantifies how much accuracy comes from instantaneous physical signals vs. contextual proxies
- Proxies: cumulative energy, battery state, GPS trajectory
- Strict mode removes proxy features for fair evaluation
-
Hybrid XGBoost + Data Reuploading (DRU) Classifier:
- Benchmarked against 5 paired nonlinear controls under identical budgets
- Controls: raw, PCA, polynomial-2, random-RBF, untrained DRU map
Key Finding
- Standalone DRU does NOT consistently match strongest classical baseline
- Trained-DRU hybrid is the only model whose mean F1 macro shifts upward from full to strict (+0.05)
- Lowest mean false-alarm rate under proxy-free evaluation
Implementation Patterns
- Use temporal partitioning (not random splits) for time-series evaluation
- Feature audit: separate physical signals from contextual proxies
- Hybrid quantum-classical: quantum feature map + classical classifier
- Evaluate under proxy-free conditions to avoid inflated metrics
- Open Qiskit 2.x implementation for reproducibility
Evaluation Protocol
- Partition by time blocks (contiguous, not random)
- Test across multiple seeds for variance estimation
- Audit features: full → loose → strict (removing proxies)
- Compare against multiple classical baselines under identical budgets
- Report inter-seed variance honestly (don't overclaim incremental gains)
Applications
- Aerospace cybersecurity analytics
- NISQ-era quantum-enhanced anomaly detection
- Cyber-physical system security monitoring
- Multi-sensor fusion evaluation
Activation
quantum anomaly detection, UAV security, data reuploading, leakage-free evaluation, temporal protocol, feature audit, Qiskit, cyber-physical systems
