Federated Learning

Federated Learning

1. Advantages

   • Less energy
   • Less wireless resources
   • Less latency
     • Somente os parâmtros do modelo são transmitidos e não os dados.
   • Data privacy
   • Higher performance
     • Mais datasets podem favorecer uma maior performance de aprendizado.

2. Types

   • FRL
     • Federated Reinforcement Learning
   • FSL
     •  federated supervised learning  
   • FL for GANS
     •  unsupervised learning  
   • FL for CL
     •  for contrastive learning (self-supervised learning)   

3. Performance Metrics

   • Delay
   • Energy
   • Reliability
   • Massive Connectivity

4. Wirless problems to Solve

   • Resource management
   • User behavior prediction
   • Channel Estimation and Signal Detection

5. Requirements

   • mURLLCs
   • Scalable architecture
   • Human-centric services