Login / Signup

A Variable Batch Size Strategy for Large Scale Distributed DNN Training.

Zhongzhe HuJunmin XiaoZhongbo TianXiaoyang ZhangHongrui ZhuChengji YaoNinghui SunGuangming Tan
Published in: ISPA/BDCloud/SocialCom/SustainCom (2019)
Keyphrases
  • batch size
  • training process
  • batch mode
  • fixed cost
  • incremental learning
  • single item
  • training data
  • objective function
  • active learning
  • semi supervised
  • supply chain
  • supervised learning
  • poisson process