MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641112745 A) filed by Chris David Raj A on September 21, 2026, for System And Method For Weightless Graph-Thermodynamic Natural Language Inference And Zero-Copy Storage Architecture.
Inventor includes Chris David Raj A.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: The invention relates to a system and method for weightless graph thermodynamic natural language inference and zero-copy storage architecture. The system eliminates reliance on parametric neural network weights and instead employs deterministic graph structures stored in a contiguous binary file format. The file comprises header, hash tables, node records, compressed sparse row adjacency arrays, and a UTF 8 string pool, all memory mapped into virtual address space without duplication into application heaps. Input text is processed into lemma or bigram stimuli, resolved into node indices using a 64 bit open addressing hash table, and propagated across adjacency arrays through graph thermodynamic physics. Energy diffusion, Hebbian decay, and bounded working memory ensure deterministic inference across conversational turns. Candidate responses are scored using a unified cognitive potential equation incorporating inhibition of return to prevent cyclic loops. The architecture supports offline deployment on consumer hardware, enabling low power execution and applicability in healthcare, defense, robotics, and sovereign personal AI systems.
Disclaimer: Curated by HT Syndication.
