MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541123229 A) filed by Vellore Institute Of Technology, Vellore, Tamil Nadu, on Dec. 6, 2025, for 'hybrid transfer learning model for text emotion detection.'
Inventor(s) include Dr Priya G; and Mr. Jeeva G.
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 hybrid transfer learning system for text emotion detection. The system includes a data preprocessing module configured to receive textual input data and generate balanced training datasets across emotion classes, a RoBERTa embedding module configured to process the textual input data using a pre-trained RoBERTa transformer model to generate contextualized sentence embeddings, a Convolutional Neural Network component configured to extract local patterns from tokenized text representations, a Bidirectional Long Short-Term Memory network component configured to capture sequential dependencies in the textual data, a feature fusion module configured to combine the contextualized sentence embeddings from the RoBERTa embedding module with feature representations from the Convolutional Neural Network component and the Bidirectional Long Short-Term Memory network component, and an output classification module configured to generate emotion class predictions based on the combined feature representations."
Disclaimer: Curated by HT Syndication.
