MUMBAI, India, Feb. 13 -- Intellectual Property India has published a patent application (202641007438 A) filed by Dmi Engineering College, Kanyakumari, Tamil Nadu, on Jan. 26, for 'deep learning system for predictive threat detection in cloud environments.'
Inventor(s) include Mrs. Monisha Raju Y V; Mrs. R. Nishanthi; Mrs. Raja Kala P; Mrs. Sheeba D; Mrs. Raju T; Mrs. J. Stephy Christina; Mrs. D. R. Sindhu; and Mrs. Dhanya Raj A.
The application for the patent was published on Feb. 13, under issue no. 07/2026.
According to the abstract released by the Intellectual Property India: "Cloud environments face escalating cybersecurity threats, including advanced persistent threats (APTs) and zero-day exploits, which traditional signature-based detection struggles to predict proactively. This invention discloses a deep learning system for predictive threat detection that leverages a hybrid convolutional neural network-long short-term memory (CNN-LSTM) architecture to analyse multi-modal data streams, such as network traffic, user behaviour logs, and resource utilization metrics, in real-time. The system preprocesses heterogeneous cloud telemetry using autoencoders for dimensionality reduction and feature extraction, followed by spatiotemporal modelling via CNN-LSTM layers to forecast anomalous patterns with high precision. Trained on augmented datasets simulating adversarial scenarios, the model achieves superior performance-up to 95% accuracy and 92% F1-score in detecting latent threats 24-48 hours in advance, as validated on benchmarks like AWS CloudTrail and Azure Sentinel logs. Integrated with auto-scaling orchestration, the system enables preemptive mitigation, reducing breach response time by 70% while minimizing false positives through attention-based interpretability mechanisms. This approach advances proactive cybersecurity in dynamic cloud infrastructures, offering scalable protection for hybrid and multi-cloud deployments."
Disclaimer: Curated by HT Syndication.
