Earticle

초록

영어

Neural networks have been reborn as a Deep Learning thanks to big data, improved processor, and some modification of training methods. Neural networks used to initialize weights in a stupid way, and to choose wrong type activation functions of non-linearity. Weight initialization contributes as a significant factor on the final quality of a network as well as its convergence rate. This paper discusses different approaches to weight initialization. MNIST dataset is used for experiments for comparing their results to find out the best technique that can be employed to achieve higher accuracy in relatively lower duration.

목차

Abstract
1. INTRODUCTION
2. MULTI-LAYER NEURAL NETWORKS TRAINING
2. 1. Structure of Multi-layer Neural Networks
2. 2. Neural Networks Training
3. WEIGHTS INITIALIZATION OF NEURAL NETWORKS
3. 1. Random Initialization
3. 2. Xavier Initialization
3. 3. He-at-al Initialization
3. 4. Batch Normal Initialization
4. EMPIRICAL RESULTS AND OBSERVATION
5. CONCLUSION
ACKNOWLEDGEMENT

키워드

저자

• Min-Jae Kang [ Professor, Department of Electronic Engineering, Jeju National University, Korea ]
• Ho-Chan Kim [ Professor, Department of Electrical Engineering, Jeju National University, Korea ] Corresponding author

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  International Journal of Advanced Culture Technology(IJACT)

• 간기

  계간

• pISSN

  2288-7202

• eISSN

  2288-7318

• 수록기간

  2013~2025

• 등재여부

  KCI 등재

• 십진분류

  KDC 600 DDC 700