Predictive Attack Path Analysis and Automated Security Prioritization for Enterprise Cloud Infrastructure Using Graph-Based Learning

Abstract

The increasing adoption of cloud computing has significantly expanded enterprise attack surfaces, creating complex security challenges involving misconfigurations, excessive permissions, vulnerable resources, and interconnected infrastructure components. Traditional vulnerability assessment approaches primarily evaluate individual security weaknesses without adequately considering the relationships between cloud assets or the potential consequences of their exploitation. This research proposes a predictive attack path analysis and automated security prioritization framework for enterprise cloud infrastructure using graph-based learning techniques. The proposed approach represents cloud environments as interconnected graphs, where nodes represent resources, identities, applications, network components, and vulnerabilities, while edges describe their relationships and access dependencies. Graph-based machine learning techniques are employed to identify potentially dangerous connectivity patterns, estimate attack path risks, and predict high-impact security exposures. Furthermore, an automated prioritization mechanism integrates vulnerability severity, resource criticality, privilege relationships, exposure levels, and potential business impact to generate context-aware remediation recommendations. The methodology aims to improve security visibility, reduce alert fatigue, and support proactive vulnerability management across dynamic cloud environments. By combining graph representation learning with predictive analytics, the proposed framework provides an intelligent approach to enterprise cloud security management, enabling security teams to prioritize remediation activities according to contextual risk rather than relying exclusively on conventional vulnerability scores.