MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541124061 A) filed by Dr. S. Sivakumar; Mr. S. Muthukumarasamy; and S. A. Engineering College, Chennai, Tamil Nadu, on Dec. 9, 2025, for 'online banking threats detection using machine learning.'
Inventor(s) include Dr. S. Sivakumar; and Mr. S. Muthukumarasamy.
The application for the patent was published on Jan. 2, under issue no. 01/2026.
According to the abstract released by the Intellectual Property India: "The invention relates to the field of cyber security for digital and online banking plat-forms and more particularly to an intelligent threat detection system capable of identify-ing and mitigating cyber-attacks in real time. The widespread adoption of digital banking and online financial services has significantly increased user convenience; however, it has also resulted in a surge of sophisticated cyber threats including phishing attacks, identity theft, unauthorized account takeovers, and large-scale fraudulent transactions. Traditional rule-based security systems are increasingly ineffective against these emerg-ing threats, as they rely on predefined patterns and lack the capability to recognize unfa-miliar or zero-day attack behaviours. This technological limitation contributes to delayed threat detection and high false-positive rates, highlighting the need for advanced data-driven security solutions. The invention introduces a machine learning-based online banking threat detection sys-tem that integrates a hybrid combination of supervised and deep learning models to clas-sify and detect malicious activities with high precision. The system incorporates Random Forest, XGBoost, LightGBM, CatBoost, Convolutional Neural Networks (CNN), and Gat-ed Recurrent Units (GRU) to perform multi-dimensional analysis of transactional data and URL patterns. By recognizing subtle distinctions between legitimate and fraudulent behaviors, the invention detects phishing attempts, abnormal financial transactions, and intrusion attempts in banking sessions. To overcome data imbalance challenges prevalent in financial cyber security datasets, the system employs the Synthetic Minority Over-sampling Technique (SMOTE), class-weight adjustments, or a combination thereof, en-suring unbiased learning and stable model performance. A continuous learning and self-adaptive model update mechanism is embedded within the system to improve its reliability against evolving cyber threats. This enables the detec-tion engine to autonomously update itself with new attack signatures, fraud trends, and user behavioural changes without manual reprogramming, thereby improving long-term scalability. Additionally, the invention includes a real-time visualization dashboard that displays detected threats, security alerts, confidence scores, and system performance met-rics, facilitating timely responses by cyber security teams. Overall, the invention provides a scalable, adaptive, and proactive cyber security frame-work for the protection of digital banking environments. Through its hybrid ML-DL ar-chitecture, adaptive learning capabilities, and real-time monitoring features, the system significantly enhances threat detection accuracy, reduces false alerts, increases opera-tional efficiency, and strengthens user trust and safety in digital financial ecosystems."
Disclaimer: Curated by HT Syndication.
