Publication Details

Abstract

This paper introduces a novel quantum-enhanced explainable AI framework for autonomous vehicle network security, featuring three synergistic innovations: a Quantum Feature Selection Module (QFSM) utilizing quantum-inspired optimization and deep SHAP analysis, a Neural Explanation Generator (NEG) employing transformer architectures, and a Cognitive Visualization Engine (CVE) integrating augmented reality with graph neural networks. The framework achieves exceptional performance with 99.5% detection accuracy and processes 220,000 packets per second while maintaining a 1.2% false alarm rate. System efficiency demonstrates 35% reduced resource utilization and 40% smaller model footprint, with rapid training cycles of 0.29 seconds and prediction speeds of 0.006 seconds. By combining quantum computing algorithms with explainable AI techniques, the framework achieves 93% threat detection accuracy, surpassing traditional methods by 6%, while maintaining 85% interpretability rates. Notable achievements include 89.7% effectiveness in identifying zero-day threats and an adaptability score of 8.9. When tested across multiple real-world autonomous driving datasets, our approach demonstrates significant improvements over conventional methods, particularly in dynamic threat environments where split-second, explainable decisions are crucial for network security. The successful integration of quantum-inspired optimization with transparent AI decision-making establishes new benchmarks in autonomous vehicle protection.