MUMBAI, India, Oct. 12 -- Intellectual Property India has published a patent application (202641118155 A) filed by Mr. Ankam Dhilli Babu; Ms. Mohammad Nourien; Mr. Bodicherla Siva Sankar; Dr. Nellore Manoj Kumar; and Dr. S. Rama on October 01, 2026, for A Computer-Implemented System And Method For Multilingual Machine Translation Into Indian Languages Using Deep Learning.
Inventors include Mr. Ankam Dhilli Babu; Ms. Mohammad Nourien; Mr. Bodicherla Siva Sankar; Dr. Nellore Manoj Kumar; and Dr. S. Rama.
The application for the patent was published on October 09, 2026, under issue no. 41/2026.
Abstract: A computer-implemented system and method for multilingual machine translation into Indian languages using deep learning are disclosed. An input module receives a source content in a source language and a selection of a target Indian language from a plurality of Indian languages, and handles code-mixed and transliterated input common in Indian-language use. A deep learning translation model, comprising a multilingual neural machine translation model such as a transformer, is configured and trained for translation into the Indian languages and addresses their specific challenges: their multiplicity and the scarcity of parallel data, by a shared multilingual model in which high-resource languages assist low-resource ones and linguistic relatedness of related languages is leveraged; their distinct scripts and rich agglutinative morphology, by a subword tokenization and a common or mapped script representation; and code-mixing and transliteration, by appropriate input handling. The model is trained by low-resource techniques, transfer learning from pretrained models, back-translation generating synthetic parallel data from monolingual data, and monolingual pretraining, mitigating data scarcity. An output module outputs the translation as text in the target script or as synthesized speech, serving communication, information access, education, commerce, and governance in the Indian languages that general systems serve poorly.
Disclaimer: Curated by HT Syndication.
