ps-CALR: Periodic-Shift Cosine Annealing Learning Rate for Deep Neural Networks

ps-CALR: Periodic-Shift Cosine Annealing Learning Rate for Deep Neural Networks

There Are Continued Efforts to Build on the Performance of Deep Learning (DL) Models in Various Fields of Application. Developing New DL Models Continues to Open Unprecedented Opportunities in Diverse Application Areas Despite the Enormous Resources Required. Generally, the Learning Mechanism of DL...

Full description

Bibliographic Details
Main Authors: Olanrewaju Victor Johnson, Chew Xinying, Khai Wah Khaw, Ming Ha Lee
Format: Article
Language:English
Published: IEEE 2023-01-01
Series:IEEE Access
Subjects:
Cosine annealing
convergence
flat minima
learning rate
loss function
optimizers
Online Access:https://ieeexplore.ieee.org/document/10348553/

Internet

https://ieeexplore.ieee.org/document/10348553/

Similar Items