Chaos-Tuned AI Promises to Predict Which Software Will Break Before It Does

Researchers in New Delhi have introduced an AI framework aimed at software maintainability prediction—estimating which parts of code are likely to become costly to fix later. The work, published in Cluster Computing by Varun Goel and Arvinder Kaur of Guru Gobind Singh Indraprastha University, claims near-99% accuracy using deep learning combined with swarm-inspired optimization and mathematical chaos theory. The pipeline targets three recurring issues in machine-learning for software data: imbalanced datasets, noisy or redundant features, and poorly tuned neural network hyperparameters. It addresses class imbalance with an improved SMOTE approach that generates more diverse synthetic minority samples, reducing distortion of the learned decision boundary. The article says maintenance can consume the largest share of a software project’s lifetime budget, and being able to flag high-maintenance modules early could enable earlier refactoring before costs escalate.