Kavach: An Intelligent AI-powered Next Gen Po...

Kavach: An Intelligent AI-powered Next Gen Polymorphic Malware Detection App

Author Name : Dr. M. K. Jayanthi Kannan, Manthan Ugale, Vivshwan Krishna Tomar, Rajat Acharya, Hitesh Kaushik, Rahul Madhuraj, Nancy Khandelwal

ABSTRACT With the proliferation of Android applications, the risk of malware has escalated, necessitating advanced detection methods. This study introduces a method that utilizes static analysis to extract comprehensive features from Android applications, including permissions, API calls, opcode sequences, services, broadcast receivers, fuzzy hashes, and file size. Two new features are also proposed to enhance detection accuracy. These features are then processed using a functional API deep learning model. The method was evaluated on a dataset comprising 14,079 samples, categorized into malware and benign applications, with malware further divided into four classes. In binary classification (malware vs. benign), the model achieved an impressive F1-score of 99.5%. For multi-class classification (five categories), the model attained an F1-score of 97%. These results demonstrate the model's superior performance compared to existing methods. The KAVACH is an Intelligent AI-powered NextGen Polymorphic Malware Detection App where Cybersecurity Meets Artificial Intelligence. The validation set is used to fine-tune hyperparameters and avoid overfitting. Model Evaluation: During validation, metrics such as precision, recall, F1-score, and confusion matrix analysis are employed to assess the model’s ability to distinguish malware from benign apps. Special emphasis is placed on reducing false negatives to ensure no malicious application is mistakenly classified as safe.