Adaptive Malware Detection and Runtime Behavioral Profiling for Securing Enterprise Cloud Workloads through Deep Learning

Abstract

The rapid adoption of cloud computing has significantly transformed enterprise IT infrastructures while introducing complex cybersecurity challenges, particularly the increasing sophistication of malware attacks targeting cloud workloads. Traditional signature-based security mechanisms often fail to identify zero-day threats, polymorphic malware, and previously unseen malicious activities. This research proposes an adaptive malware detection and runtime behavioral profiling framework using deep learning techniques to strengthen enterprise cloud workload security. The proposed approach integrates machine learning-based threat classification, continuous runtime monitoring, behavioral analysis, and adaptive anomaly detection to identify malicious activities within dynamic cloud environments. Runtime behavioral profiling examines system calls, network traffic, process execution patterns, resource utilization, and application interactions to distinguish legitimate workload behavior from suspicious activities. Deep learning architectures, including Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and hybrid models, are considered for extracting complex behavioral patterns and improving malware classification accuracy. The framework also incorporates continuous model adaptation to address evolving attack techniques and changing workload characteristics. The proposed research aims to improve detection accuracy, reduce false-positive rates, strengthen real-time threat identification, and enhance enterprise cloud resilience. By combining behavioral intelligence with adaptive deep learning, the framework provides a scalable and proactive approach to protecting modern enterprise cloud infrastructures against emerging malware threats.