Streamlining Compliance: Orchestrating Automated Checks for Cloud-based AI/ML Workflows

Abstract

Ensuring security and safeguarding data privacy within cloud workflows has garnered considerable attention in research circles. For instance, protecting the confidentiality of patients' private data managed within a cloud-deployed workflow is crucial, as is ensuring secure communication of such sensitive information among various stakeholders. In light of this, our paper proposes an architecture and a formal model for enforcing security within cloud workflow orchestration. The proposed architecture underscores the importance of monitoring cloud resources, workflow tasks, and data to identify and anticipate anomalies in cloud workflow orchestration. To achieve this, we advocate a multi-modal approach combining deep learning, one-class classification, and clustering techniques. In summary, our proposed architecture offers a comprehensive solution to security enforcement within cloud workflow orchestration, leveraging advanced techniques like deep learning for anomaly detection and prediction, particularly pertinent in critical domains such as healthcare during unprecedented times like the COVID-19 pandemic.