Anomaly Detection in Cloud API Access PatternsUsing Temporal Convolutional Networks

Abstract

Cloud platforms generate massive API access logs, where abnormal patterns may indicate security breaches,insider threats, or compromised credentials, demanding intelligent and automated anomaly detection mechanisms.Conventional approaches employ segmentation, statistical profiling, clustering, recurrent networks, and supervisedclassifiers to model sequential API behavior and distinguish normal activities from malicious deviations. Thesetechniques generally achieve high detection accuracy and improved threat visibility, enhancing cybersecuritymonitoring systems while reducing manual auditing efforts in large-scale distributed cloud environments. However,they struggle with evolving attack patterns, high false-positive rates, limited temporal dependency modelling, dataimbalance, and poor generalization across heterogeneous cloud infrastructures. This study proposes a self-supervisedTemporal Convolutional Network with adaptive anomaly scoring, achieving robust sequential modelling, reduced falsealarms, and improved detection stability in cloud APIs. A self-supervised Temporal Convolutional Network modelssequential API behavior using causal dilated convolutions and adaptive scoring, enabling accurate, scalable, and realtime cloud anomaly detection.