An Efficient Dropout for Robust Deep Neural Networks

Overfitting remains a major difficulty in training deep neural networks, especially when attempting to achieve good generalization in complex classification tasks. Standard dropout is often employed to address this issue; however, its uniform random inactivation of neurons typically leads to instabi...

Full description

Saved in:
Bibliographic Details
Main Authors: Yavuz Çapkan, Aydın Yeşildirek
Format: Article
Language:English
Published: MDPI AG 2025-07-01
Series:Applied Sciences
Subjects:
Online Access:https://www.mdpi.com/2076-3417/15/15/8301
Tags: Add Tag
No Tags, Be the first to tag this record!