Concepedia

Weight Initialization

Weight Initialization is a method in neural network training that specifies the initial values of the model’s learnable parameters before optimization begins, distinguishing itself from other initialization techniques by setting the starting point of the parameters to influence convergence and performance.

330

Publications

37.1K

Citations

973

Authors

397

Institutions

Publications per year

2017–2026

180

Parents

Publications

330

Understanding the difficulty of training deep feedforward neural networks

Xavier Glorot, Yoshua Bengio · 2010 · 12.7K citations

, , +13

Learning and development in neural networks: the importance of starting small

Jeffrey L. Elman · Cognition · 1993 · 2K citations

, , +8

Interpreting neural-network connection weights

G. David Garson · AI Expert archive · 1991 · 1.5K citations

, , +8

Improving the learning speed of 2-layer neural networks by choosing initial values of the adaptive weights

Dat Nguyen, Bernard Widrow · 1990 · 1.4K citations

, , +13

Evolving Deep Convolutional Neural Networks for Image Classification

Yanan Sun, Bing Xue, Mengjie Zhang et al. · IEEE Transactions on Evolutionary Computation · 2019 · 736 citations

, , +15

1–5 of 330

Authors

973

  1. PC
    Pravin Chandra

    Guru Gobind Singh Indraprastha University, India

    5 publications · 87 citations

    5

    h-index

  2. YB
    Yoshua Bengio

    Université de Montréal, Canada

    4 publications · 12.9K citations

    4

    h-index

  3. SF
    Syed Furqan Qadri

    Shenzhen University, China

    3 publications · 161 citations

    3

    h-index

  4. IE
    Ioannis E. Livieris

    University of Patras, Greece

    3 publications · 79 citations

    3

    h-index

  5. SS
    Samuel S. Schoenholz

    Google (United States), United States

    3 publications · 198 citations

    3

    h-index

1–5 of 973

Institutions

397

  1. Google (United States)

    Mountain View, United States

    7 h-index

    11 publications

    1K citations

  2. Beijing University of Technology

    Beijing, China

    6 h-index

    19 publications

    1.4K citations

  3. Guru Gobind Singh Indraprastha University

    Delhi, India

    5 h-index

    8 publications

    152 citations

  4. Shenzhen University

    Shenzhen, China

    4 h-index

    14 publications

    358 citations

  5. Princeton University

    Princeton, United States

    4 h-index

    7 publications

    377 citations

  6. City University of Hong Kong

    Hong Kong, Hong Kong

    4 h-index

    10 publications

    677 citations

  7. University of Technology Sydney

    Sydney, Australia

    4 h-index

    8 publications

    1.3K citations

  8. Chinese Academy of Sciences

    Beijing, China

    4 h-index

    4 publications

    1K citations

1–8 of 397

Venues

  1. arXiv (Cornell University)

    Repository · 56 publications · 4.1K citations

    27

    h-index

  2. Neurocomputing

    14 publications · 733 citations

    13

    h-index

  3. IEEE Transactions on Neural Networks

    12 publications · 947 citations

    12

    h-index

  4. Neural Information Processing Systems

    Conference · 8 publications · 549 citations

    8

    h-index

  5. Neural Networks

    8 publications · 202 citations

    8

    h-index

1–5