Keynote Talk 1: Efficient DNN Training at Scale: from Algorithms to Hardware.
Gennady PekhimenkoPublished in: IPDPS Workshops (2022)
Keyphrases
- computationally expensive
- parallel architectures
- computationally complex
- computationally efficient
- computationally intensive
- efficient implementation
- algorithmic solutions
- data structure
- low cost
- highly scalable
- parallel hardware
- real time
- computationally demanding
- pruning strategy
- highly efficient
- optimization problems
- hardware and software
- times faster
- training process
- orders of magnitude
- massively parallel
- computer systems
- efficient solutions
- worst case
- significant improvement
- multiscale
- machine learning
- neural network