MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202641098915 A) filed by Vellore Institute Of Technology on August 15, 2026, for Multi Metric Behavioral Drift Detection For Identification Of Non-Observable Failures In Web Applications.
Inventors include R Sujatha; Darsani R; and Ponmagal H.
The application for the patent was published on August 21, 2026, under issue no. 34/2026.
Abstract: “Multi Metric Behavioral Drift Detection for Identification of Non-Observable Failures in Web Applications” An adaptive multi-metric behavioral drift detection system and method for early identification of non-observable failures in web applications are disclosed. The system comprises a metric collection module to gather real- time performance metrics, a baseline modeling module to establish and dynamically update a statistical baseline model of normal behavior, and a drift detection module to calculate deviations representing behavioral drift. A multi-metric correlation module evaluates relationships among metrics to identify compound abnormal patterns. A composite risk scoring module aggregates the deviations into a single unified risk score based on deviation magnitude and operational impact. An adaptive threshold module dynamically recalibrates alert thresholds based on recent system volatility and workload patterns. An alert and visualization module generates real-time notifications when the composite risk score exceeds the recalibrated threshold, enabling proactive corrective actions before service disruption occurs. [FIG. 2]
Disclaimer: Curated by HT Syndication.
