MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641107294 A) filed by Kkr & Ksr Institute Of Technology And Sciences on September 07, 2026, for Self-Evolving Neural Networks Using Performance Driven Architecture Mutation.
Inventors include Mrs. Dr. P. G. K. Sirisha; Mr. Prof. Repudi Ramesh; Achanti Navya; Gurivisetti Varshini; Gajula Saranya; Chinta Bhavana; and Desavarapu Vindu.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: ABSTRACT [0010] A self-evolving neural network framework is disclosed that autonomously adapts its architecture through performance-driven mutation rather than relying on predetermined layer counts, neuron numbers, or connection patterns. A population of initial network structures is evaluated against metrics such as accuracy, loss, computational efficiency and model complexity. Higher-fitness architectures are selected for reproduction, while underperforming ones are discarded. Evolutionary operators including mutation and crossover then generate new candidates by adding or removing layers, altering neuron counts, modifying connections, or changing activation functions. Successive generations produce progressively more efficient and task-specific networks. The resulting system reduces human intervention, improves adaptability to changing problem domains and yields robust learning models suitable for classification, regression, optimization, and decision- making tasks.
Disclaimer: Curated by HT Syndication.
