BROWSE
Detail
Conference Paper
Deep Learning Model Validation Method based on the Shared Memory in Distributed High Performance Computing
Cited - time in 
Share 
Share
- Authors
- Eun-Ji Lim, Shin-Young Ahn, Wan Choi
- Issue Date
- 2019-12
- Citation
- International Conference on Internet (ICONI) 2019, pp.384-385
- Language
- English
- Type
- Conference Paper
- Abstract
- This paper presents a new deep learning model validation and training method to improve the reliability of the validation result and reduce the model training time. The proposed method enables distributed processes run model training and validation in parallel by sharing model parameters through high-speed remote shared memory in distributed computing environments. We evaluated our method by training and validating the Inception-v3 model with ImageNet dataset and showed that our method enables more precise model validation while reducing model training time.
- KSP Keywords
- Distributed processes, High Speed, High-performance computing(HPC), Inception-v3, Model Validation, Model parameter, Precise model, Shared Memory, Sharing model, Training time, deep learning(DL)