Outsourcing secured machine learning (ML)-as-a-service for causal impact analytics in multi-tenant public cloud

Public cloud platform provides multi-tenant software leased services. Therefore, various participants can use software agents to effectively conduct their services in an automated machine learning pipeline. We will show how to construct numerous software-as-a-service (SaaS) service modules on a platform-as-a-service (PaaS) delivered by a well-known cloud service provider to achieve secured big data causal impact analytics. Given a secured cloud provider, it is still considered as an honest-but-curious adversary. Similar assumption will also be applied to other participants, including data brokers, data users, causal impact model builders and service users. These participants will leverage their software agents from SaaS perspectives to provide machine learning pipeline services. The greatest research challenge is how to integrate three types of SaaS, i.e. security-as-a-service (Sec-as-a-Service), machine learning-as-a-service (ML-as-a-Service) and data broker-as-a-service (DB-as-a-Service), to achieve secured causal impact analytics in the public cloud. We will establish three types of SaaS on top of AWS or GCP. We assume a machine learning causal impact model and data are all protected and confidential when we proceed a machine learning model's training and online testing for causal impact analytics. We will select a machine learning algorithm, and integrate it into privacy and secured data protection algorithms and further enable a two-phase machine learning process for a causal-impact model's training and online testing. Furthermore, we will investigate how to effectively embed three types of SaaS into automated machine learning pipeline and achieve secured machine learning for causal impact analytics in the public cloud.