MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202611117195 A) filed by Sharda University on September 30, 2026, for Hybrid Multi-Model Network Intrusion Detection System With Explainability And Evidence- Qualified Threat Mapping.
Inventors include Ansh Pasnavat; Deepak Maindola; and Kapil Kumar.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: A hybrid network intrusion detection system (1000) transforms flow records into processed feature vectors by a preprocessing module (200), processes each processed feature vector by a reconstruction-based anomaly module (300) to obtain a reconstruction-error value (340) and by a partitioning-based anomaly module (400), and determines an anomaly indication by a logical disjunction of anomaly flags. A hybrid decision engine (600) deterministically routes each flow record to one of a first decision path (620) applying a supervised classification module (500) and a second decision path (630) applying a session-level sequence-verification module (700), exactly one being applied per flow record, and outputs a decision status with a confidence value. The supervised classification module (500) may be augmented with a latent representation (320) and the reconstruction-error value (340) appended as scalar columns, a variant being selected by a comparative metric. An explainability module (800) and a mapping module (900) provide interpretable, evidence-qualified outputs. Fig. 1
Disclaimer: Curated by HT Syndication.
