Capuchin: Tensor-based GPU Memory Management for Deep Learning.
Xuan PengXuanhua ShiHulin DaiHai JinWeiliang MaQian XiongFan YangXuehai QianPublished in: ASPLOS (2020)
Keyphrases
- deep learning
- memory management
- parallel computation
- operating system
- unsupervised learning
- hardware implementation
- real time
- parallel computing
- machine learning
- parallel algorithm
- unsupervised feature learning
- parallel implementation
- mental models
- graphics processing units
- parallel processing
- weakly supervised
- image processing
- flash memory
- pairwise
- active learning