Non-iid data and Continual Learning processes in Federated Learning: a long road ahead

Document Outline

• Non-IID data and Continual Learning processes in Federated Learning: A long road ahead
  • Introduction
  • Overview on Federated Learning
  • Non-IID data in Federated Learning
    • Taxonomy of data heterogeneity
    • State-of-the-art classification: Personalization strategies
      • Client-level personalization
      • Group-level personalization
    • Statistical taxonomy for non-IID strategies
      • Changes in the input space throughout clients
      • Changes in the behaviour throughout clients
  • Non-IID Data in Continual Learning: Concept Drift
    • Concept drift definition
    • Concept drift detection
  • Addressing Federated and Continual non-IID data
    • Virtual concept drifts
    • Real concept drifts
    • Data requirements for the different scenarios
  • Experiments
    • Spatial non-IID scenarios
    • Temporal non-IID scenarios
  • Challenges and future directions
  • CRediT authorship contribution statement
  • Declaration of competing interest
  • Acknowledgements
  • References