Lottery Hypothesis based Unsupervised Pre-training for Model Compression in Federated Learning.
Sohei ItaharaTakayuki NishioMasahiro MorikuraKoji YamamotoPublished in: VTC Fall (2020)
Keyphrases
- supervised learning
- learning algorithm
- prior knowledge
- learning process
- probabilistic model
- computational model
- unsupervised learning
- learning scheme
- objective function
- learning systems
- high level
- structured prediction
- em algorithm
- online learning
- image compression
- generative model
- parameter estimation
- image quality
- mathematical model
- learning tasks
- data compression
- neural nets
- reinforcement learning
- decision theoretic
- weakly supervised
- similarity measure