[10p-E219-1~10] 4.7 Quantum Optics, Nonlinear Optics and Structured Optics

15:45 〜 16:00英語発表

[10p-E219-6] Interpretable Machine Learning for Secure Key Rate Prediction and Fiber Anomaly Detection in Twin-Field QKD over Multicore Fiber

〇(D)Neha Kumari¹, Ritu Raj Singh¹ (1.Netaji Subhas University of Technology)

キーワード：

Twin-Field Quantum Key Distribution、Multicore Fiber、Quantum Communication

Interpretable machine learning predicts secure key rate and detects fiber anomalies in multicore-fiber Twin-Field QKD (TF-QKD). Trained in 3,000 simulations, XGBoost achieved R²=0.951 and RMSE=0.044 Mbps for secure key rate estimation. SHAP identified classical power and distance as dominant factors. Isolation Forest detected anomalous fiber states with 0.87 precision and 0.83 recall.