pFedMoE: Data-Level Personalization with Mixture of Experts for Model-Heterogeneous Personalized Federated Learning.
Liping YiHan YuChao RenHeng ZhangGang WangXiaoguang LiuXiaoxiao LiPublished in: CoRR (2024)
Keyphrases
- prior knowledge
- learning models
- learned models
- human experts
- learning process
- learning scheme
- probabilistic model
- hidden variables
- database
- input data
- data sources
- data analysis
- sensory data
- training data
- test data
- experimental data
- data sets
- domain experts
- statistical methods
- missing data
- network structure
- accurate models
- user model
- reinforcement learning
- active learning
- learning environment
- bayesian methods
- personalized learning
- dirichlet process
- discrete data
- latent variable models
- data structure
- probability distribution
- gaussian distribution
- data points
- supervised learning
- mixture model