Context Aware AI for Multi-Modal Fraud Detection Using IP Pattern and Human Interaction Behavior
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Financial fraud continues to threaten digital banking and online payment systems, particularly through device misuse, IP manipulation, and identity-related inconsistencies.Existing detection methods, including blacklist-based systems, clustering algorithms like DBSCAN, and optimization techniques such as genetic algorithms, have achieved limited success.They often rely on static rules, struggle with evolving fraud patterns, and produce high false alarm rates, making realtime detection in large-scale financial environments challenging.This research proposes an AI-driven fraud detection framework using deep learning models-BERT, Multi-Layer Perceptron (MLP), and Autoencoders-to analyze device usage patterns, IP anomalies, and KYC irregularities.BERT captures textual and sequential identity features, MLP classifies accounts as fraudulent or legitimate, and Autoencoders detect anomalies by reconstructing normal behavioral profiles and flagging deviations.Unlike clustering or optimization-based approaches, this framework learns directly from raw behavioral and contextual features, enabling adapt ability to novel fraud strategies.The system is scalable, real-time, and reduces false positives, offering financial institutions a robust solution against identity forgery, device/IP manipulation, and evolving fraud behaviours.
Fields of Study & Subjects
Imbalanced Data Classification Techniques
Spam and Phishing Detection
User Authentication and Security Systems
Publication details
- Journal / source
- Advances in intelligent systems research/Advances in Intelligent Systems Research
- Pages
- 612–627
- Publication date
- 1 April 2026
- Language
- EN
- Document type
- Book chapter