MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202641092314 A) filed by Vardhaman College Of Engineering on July 30, 2026, for A System And Method For Ai-Driven Software Defect Prediction And Automated Code Optimization.
Inventors include Dr. Reddy Saisindhutheja; Mr. Tummala Vinay; Ms. Monika Garg; Dr. Gagandeep Arora; Mr. Manish Chhabra; and Ms. Baru Prashanthi.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: ABSTRACT A System and Method for AI-Driven Software Defect Prediction and Automated Code Optimization The present disclosure relates to a system and method for software defect prediction and runtime-verified automated code optimization. The system collects source-code artifacts, executable traces, workload characteristics and hardware-performance telemetry and temporally aligns software events with processor, cache, memory, storage and network events. The processor constructs a workload-conditioned causal execution graph and calculates defect-risk, optimization-suitability and causal-responsibility values for code regions . An impact boundary is defined for a code region of interest and a constrained transformation engine produces a candidate code modification within functional, security and resource boundaries. The original and modified versions of the code are executed under the same conditions in an isolated validation environment. The system checks for equivalence and compares processor cycles, cache behaviour, memory consumption, execution latency and energy usage. A change is rolled out progressively only if the defect risk is reduced and a resource-improvement threshold is satisfied. Runtime-drift monitoring and reversible checkpoints enable recovery while verified results update prediction and optimization parameters.
Disclaimer: Curated by HT Syndication.
