GeePS: scalable deep learning on distributed GPUs with a GPU-specialized parameter server.
Henggang CuiHao ZhangGregory R. GangerPhillip B. GibbonsEric P. XingPublished in: EuroSys (2016)
Keyphrases
- deep learning
- commodity hardware
- graphics processing units
- map reduce
- general purpose
- parallel processing
- parallel computing
- parallel computation
- gpu implementation
- graphics hardware
- unsupervised learning
- cloud computing
- unsupervised feature learning
- parallel programming
- mental models
- graphics processors
- deep architectures
- machine learning
- computing systems
- memory bandwidth
- higher order
- semi supervised
- parallel implementation
- weakly supervised
- decision support
- domain specific
- supervised learning