Hierarchical Distributed-Memory Multi-Leader MPI-Allreduce for Deep Learning Workloads.
Truong Thao NguyenMohamed WahibRyousei TakanoPublished in: CANDAR Workshops (2018)
Keyphrases
- distributed memory
- deep learning
- shared memory
- parallel implementation
- unsupervised learning
- restricted boltzmann machine
- deep belief networks
- unsupervised feature learning
- parallel computers
- ibm sp
- parallel architecture
- machine learning
- parallel machines
- message passing
- weakly supervised
- parallel computing
- mental models
- parallel algorithm
- multithreading
- computer systems
- feature space
- image processing