MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641108767 A) filed by Indian Institute Of Information Technology, Design And Manufacturing, Kancheepuram Iiitdm on September 10, 2026, for System And Method For Failure-Oriented Instructional Code Synthesis Using Multi-Objective Generative Reasoning.
Inventors include Shak Noor Mahammad; Dhayalakumar M; Mohit Mahendra Bhasme; and Elanthiryan S.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: The present invention relates to computer-implemented generation of software code. A software-generation objective is received and at least a portion of a software implementation is generated and represented in a machine-readable form, such as a code or capability graph. A Behavioural Instability Analysis Engine analyses the representation to identify a behavioural deviation associated with a condition of the generated implementation. A Behavioural Instability Representation is generated based on the identified deviation and is provided to a Generation-Steering Module. The Generation-Steering Module uses the representation to select, modify or regenerate a subsequent portion of the software implementation before completion. In one embodiment, graph-based analysis including Laplacian-based anomaly detection is used to identify the behavioural deviation. The system may be used in an educational environment for generating programming exercises, examples or instructional code, thereby assisting a tutor in obtaining behaviourally reliable generated code with reduced manual checking.
Disclaimer: Curated by HT Syndication.
