DETECTION OF SCHEMA TO IDENTIFY CYBER THREATS BY USING AI

Abstract

A Novel AI-Driven Schema Detection Framework for Cyber Threat Identification In the evolving landscape of cyber security, traditional signature-based detection methods fall short against sophisticated, polymorphic threats such as zero-day exploits and advanced persistent threats (APTs). This paper proposes a survey on AI-driven framework for schema detection in cyber threat identification, leveraging hybrid schema-on-read and schema-on-write data management approaches integrated with deep learning models. Our approach dynamically infers and validates data schemas from heterogeneous network logs, endpoint behaviors, and unstructured threat intelligence feeds, enabling real-time anomaly detection without predefined rules. The framework lies in its adaptive schema evolution mechanism, which employs reinforcement learning to iteratively refine schema definitions based on emerging threat patterns, achieving a 25% improvement in detection accuracy over state-of the-art methods like those using static schema-on-write normalization. By incorporating natural language processing (NLP) for semantic schema matching and graph neural networks (GNNs) for relational threat mapping, the system not only identifies known malware variants but also predicts novel attack schemas through unsupervised clustering of behavioral anomalies. XAI (Explainable AI) AI systems designed to provide transparent and interpretable decision-making. federated learning, and robust detection for edge/IoT environments.