Reducing the Training Overhead of the HPC Compression Autoencoder via Dataset Proportioning.
Tong LiuShakeel AlibhaiJinzhen WangQing LiuXubin HePublished in: NAS (2021)
Keyphrases
- training dataset
- restricted boltzmann machine
- feedforward neural networks
- image compression
- training samples
- training set
- compression algorithm
- compression scheme
- data compression
- compression ratio
- benchmark datasets
- data sets
- synthetic datasets
- training process
- hidden markov models
- supervised learning
- fault tolerance
- training examples
- database
- video sequences
- training algorithm
- fault tolerant
- learning algorithm
- training phase
- multi layer perceptron
- high performance computing
- neural network
- compression rate