D-Cliques: Compensating for Data Heterogeneity with Topology in Decentralized Federated Learning.
Aurélien BelletAnne-Marie KermarrecErick LavoiePublished in: SRDS (2022)
Keyphrases
- data collection
- data sets
- data sources
- data analysis
- learning systems
- training data
- prior knowledge
- data quality
- learning algorithm
- statistical analysis
- distributed systems
- input data
- database
- synthetic data
- data processing
- raw data
- learning models
- data structure
- high quality
- data points
- online learning
- original data
- search engine
- database systems
- missing data
- background knowledge
- databases
- knowledge discovery
- neural network
- human experts
- data distribution
- experimental data
- reinforcement learning
- computer systems
- active learning
- end users
- image data