Yet Another Accelerated SGD: ResNet-50 Training on ImageNet in 74.7 seconds.
Masafumi YamazakiAkihiko KasagiAkihiro TabuchiTakumi HondaMasahiro MiwaNaoto FukumotoTsuguchika TabaruAtsushi IkeKohta NakashimaPublished in: CoRR (2019)
Keyphrases
- stochastic gradient descent
- linear svm
- training examples
- training set
- training process
- response time
- classifier training
- training speed
- training phase
- training algorithm
- test set
- supervised learning
- data sets
- image database
- collaborative filtering
- object detection
- object recognition
- feature selection
- learning algorithm