• Danilo Ardagna (POLIMI), Ignacio Blanquer, Miguel Caballer, Amanda Calatrava (UPV) and Marco Vieira (UC)
  • Francisco Brasileiro, Reinaldo Gomes (UFCG) and Wagner Meira Jr. (UFMG) - Brazil

Trustworthiness in data analytics and cloud services comprises a broad range of properties. Dimensions such as Security, Privacy protection, Quality of Services, Reliability, and Fairness impose tackling a set of challenges at different layers. ATMOSPHERE has designed a software architecture to provide performance simulation, Dynamic Quality of Service adaptation, vulnerability assessment, privacy leakage risk estimation and model neutrality evaluation on top of a federated cloud infrastructure. The platform supports composing data analytics workflows, defining QoS and privacy restrictions (such as using enclaves), ethical targets (for evaluating potential discrimination bias of models) and executing such workflows on top of a selfmanaged cloud platform.