MUMBAI, India, Sept. 7 -- Intellectual Property India has published a patent application (202641059994 A) filed by Hindustan Institute Of Technology And Science on May 12, 2026, for Adaptive Federated Learning System For Privacy-Preserving Cyber Intrusion Detection Across Distributed Networks.
Inventors include Sharmila E; and Dr. Sudalaimuthu T.
The application for the patent was published on September 04, 2026, under issue no. 36/2026.
Abstract: The present invention generally relates to the field of cybersecurity, artificial intelligence, distributed computing, and machine learning. More particularly, the present invention relates to an adaptive federated learning system for privacy-preserving cyber intrusion detection across distributed networks, wherein cyber intrusion detection models are collaboratively trained across distributed environments without sharing raw cybersecurity data. With the rapid expansion of cloud computing, Internet of Things (IoT), enterprise systems, and distributed digital infrastructures, cyber threats have become increasingly sophisticated and frequent. Traditional intrusion detection systems rely upon centralized collection of network traffic, endpoint logs, and cybersecurity information for attack analysis and model training. However, centralized cybersecurity systems suffer from limitations such as privacy risks, excessive computational burden, high communication overhead, delayed detection response, and vulnerability to single-point failures. Transmission of raw cybersecurity data to centralized servers also raises concerns related to data confidentiality, unauthorized access, and regulatory compliance. Federated learning has emerged as a distributed machine learning approach wherein multiple entities collaboratively train intelligent models without transferring raw data. Although federated learning improves privacy preservation, existing intrusion detection frameworks are often limited by static learning approaches, poor adaptability to evolving cyber threats, insufficient privacy-preserving mechanisms, and inefficient model aggregation methods. Therefore, there exists a need for an adaptive federated learning system capable of providing privacy-preserving, scalable, intelligent, and real-time cyber intrusion detection across distributed networks while ensuring confidentiality of cybersecurity information.
Disclaimer: Curated by HT Syndication.
