Intellectual Property India Publishes Patent Application for 'Transformer-Based Malware Detection System With Explainable AI' Filed by Vellore Institute Of Technology

MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541123245 A) filed by Vellore Institute Of Technology, Vellore, Tamil Nadu, on Dec. 6, 2025, for 'transformer-based malware detection system with explainable ai.'

Inventor(s) include Dr. Aju D; Ms. Nabiha Younus Khan; and Ms. Niranjana Boby.

The application for the patent was published on Jan. 2, under issue no. 01/2026.

According to the abstract released by the Intellectual Property India: "The present disclosure provides a malware detection system including a data processing module configured to receive portable executable files and extract static features therefrom, a transformer-based neural network configured to receive the extracted static features and generate malware classification outputs, and an explainable artificial intelligence module integrated with the transformer-based neural network and configured to generate interpretability outputs for the classification results. The data processing module extracts n-gram byte patterns from hexadecimal representations of the portable executable files through a FeatureExtraction module that processes pairs of adjacent hexadecimal values to identify recurring byte patterns. The system includes a preprocessing unit that applies normalization and implements class weighting techniques to address imbalanced distributions. The explainable artificial intelligence module incorporates a SHAPexplain module for global feature importance analysis and a LIMEExplain module for instance-specific explanations, generating visual representations that enable cybersecurity analysts to understand and validate automated threat detection results."

Disclaimer: Curated by HT Syndication.