MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641100007 A) filed by Dr Yalakala Dinesh Kumar; Ayyagari Prasada Rao; Manda Sravanti; Killugu Eswar Anil Kumar Reddy; Singamsetti Jhansi Lakshmi; and Gudivada Prasad on August 19, 2026, for Self-Evolving Neural Intelligence Framework With Continuous Knowledge Refinement For Predictive Analytics.

Inventors include Dr Yalakala Dinesh Kumar; Ayyagari Prasada Rao; Manda Sravanti; Killugu Eswar Anil Kumar Reddy; Singamsetti Jhansi Lakshmi; and Gudivada Prasad.

The application for the patent was published on August 28, 2026, under issue no. 35/2026.

Abstract: Self-Evolving Neural Intelligence Framework with Continuous Knowledge Refinement for Predictive Analytics ABSTRACT The present invention discloses a self-evolving neural intelligence framework for continuous knowledge refinement and predictive analytics. The framework is configured to acquire data from dynamic and heterogeneous sources, preprocess and contextualize the received data, and generate predictive outputs using one or more adaptive neural models. A continuous knowledge refinement module evaluates newly received information based on relevance, novelty, prediction error, confidence, statistical deviation, and data quality, and selectively updates a knowledge repository containing learned representations. A continuous learning mechanism incrementally modifies model parameters, neural pathways, learning characteristics, or architecture without requiring complete retraining. A knowledge retention mechanism preserves significant previously learned information and mitigates catastrophic forgetting during adaptation. The framework further includes drift detection, feedback evaluation, performance monitoring, and adaptive governance mechanisms for detecting changing data patterns, validating model updates, and controlling autonomous self-evolution. Predictive outcomes are compared with subsequently observed results to generate feedback signals for iterative refinement.

Disclaimer: Curated by HT Syndication.