High-precision malware detection in android apps using quantum explainable hierarchical interaction network

The exponential growth of Android applications has increased the prevalence of fraudulent and malicious apps, posing significant risks to user security and privacy. Existing detection methodologies often struggle with poor interpretability, scalability, and computational complexity, limiting their effectiveness. To address these challenges, this study introduces the Quantum Explainable Hierarchical Interaction Network (QEHIN), a novel framework designed to detect real and fake Android applications with superior accuracy and interpretability. QEHIN incorporates quantum computing principles such as superposition and entanglement to model high-order feature interactions effectively. Its innovative architecture includes a Quantum Embedding Layer for transforming input features into quantum states, a Quantum Hierarchical Interaction Network (QHIN) for capturing complex dependencies, a Quantum Deep Neural Network (QDNN) for enhanced feature processing, and a Quantum Cross-Hierarchical Unit (QCHU) to ensure seamless integration across hierarchical levels. This design achieves precise, transparent, and scalable detection of malicious applications, addressing the shortcomings of traditional methods. Evaluation on the Google Play Store Reviews, MobileRec, and Android-AppRecommendation datasets demonstrates the novelty and effectiveness of QEHIN. It achieves an accuracy of 98.86 %, precision of 98.78 %, recall of 98.82 %, and a kappa score of 98.54 %, significantly outperforming existing approaches.