MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202644088030 A) filed by C Pavani; Chandrashekhar; Priya Gupta; Mukesh Kamath Bola; Javid Iqbal Thirupattur; Shaik Shabana Anjum; and Dayananda Sagar Academy Of Technology And Management on July 18, 2026, for Hybrid Explainable Machine Learning Framework For Synthetic Media Detection.
Inventors include C Pavani; Chandrashekhar; Priya Gupta; Mukesh Kamath Bola; Javid Iqbal Thirupattur; Shaik Shabana Anjum; and Dayananda Sagar Academy Of Technology And Management.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: The present invention relates to a Hybrid Explainable Machine Learning Framework for Synthetic Media Detection, designed to accurately identify AI-generated images, videos, and audio while providing transparent and interpretable detection results. The framework integrates deep learning, conventional machine learning, and Explainable Artificial Intelligence (XAI) techniques to improve the reliability and trustworthiness of synthetic media detection. The system extracts spatial, temporal, and spectral features from multimedia content using convolutional neural networks, transformer-based models, and handcrafted feature extraction methods. These features are fused and processed through a hybrid classification engine to distinguish authentic media from manipulated or AI-generated content. Explainability modules, including feature attribution and visual attention mapping, generate human-interpretable explanations highlighting the regions and characteristics influencing the detection decision. The framework further incorporates continuous learning, confidence estimation, and cloud-based analytics to adapt to emerging deepfake generation techniques. An interactive dashboard provides real-time detection reports, visual explanations, confidence scores, and forensic insights for investigators, content moderators, media organizations, and cybersecurity professionals. The proposed invention enhances detection accuracy, model transparency, user trust, and adaptability, thereby providing a comprehensive and scalable solution for combating synthetic media, misinformation, identity fraud, and digital content manipulation.
Disclaimer: Curated by HT Syndication.
