AdaComp : Adaptive Residual Gradient Compression for Data-Parallel Distributed Training.
Chia-Yu ChenJungwook ChoiDaniel BrandAnkur AgrawalWei ZhangKailash GopalakrishnanPublished in: AAAI (2018)
Keyphrases
- data sets
- database
- data collection
- data analysis
- image data
- knowledge discovery
- parallel distributed
- supervised learning
- small number
- learning algorithm
- labelled data
- high quality
- original data
- raw data
- missing data
- data processing
- training data
- data reduction
- synthetic data
- training examples
- data sources
- image compression
- input data
- simulated annealing
- data structure
- wavelet transform
- evolutionary algorithm