Design of a Quantization-Based DNN Delta Compression Framework for Model Snapshots and Federated Learning.
Haoyu JinDonglei WuShuyu ZhangXiangyu ZouSian JinDingwen TaoQing LiaoWen XiaPublished in: IEEE Trans. Parallel Distributed Syst. (2023)
Keyphrases
- learning scheme
- conceptual framework
- conceptual model
- probabilistic model
- prior knowledge
- learning algorithm
- learning models
- mathematical model
- learning mechanism
- formal model
- computational model
- dynamic bayesian networks
- learning systems
- learned models
- inference process
- bayesian framework
- modelling language
- modeling framework
- active learning
- learning process
- context dependent
- distributed cognition
- simulation environment
- machine learning
- cognitive theory
- object detectors
- trust model
- metamodel
- design principles
- theoretical framework
- em algorithm
- data sources
- reinforcement learning
- case study