MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641104732 A) filed by Vardhaman College Of Engineering on August 31, 2026, for Multimodal Emotion Recognition Using Cross-Attention Mechanisms In Human-Al Interaction.
Inventors include Dr. Ganesh Bhaiyya Regulwar; Ms. T. Prashanthi; Mr. K. Venkatesh; Dr. K. Nikhila; Dr. L. Sunitha; and Dr. E. Ravi Kumar.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: Multimodal Emotion Recognition Using Cross-Attention Mechanisms in Human-Al Interaction is the proposed invention. The proposed invention integrates facial expressions, speech signals, textual information, body gestures, and physiological measures to generate rich representations of emotion. Dedicated deep neural networks extract modality-specific features, and a cross-attention fusion architecture dynamically learns the contextual relationship among multiple modalities by focusing on the most informative emotional cues. Modelling of temporal sequences captures the changing emotional patterns in continuous interactions and adaptive personalisation enhances the recognition performance for individual users. The framework also incorporates explainable artificial intelligence, privacy-preserving processing and fairness-aware optimisation to ensure that the operation is transparent, secure and reliable. The invention can more effectively achieve the effect of emotion recognition by intelligently modelling the cross-modal dependencies rather than the traditional feature fusion method, thereby improving the accuracy, robustness, adaptability, and real-time performance. The proposed invention can be applied to virtual assistants, intelligent education, healthcare monitoring, customer service automation, social robotics, and next-generation affective computing systems.
Disclaimer: Curated by HT Syndication.
