AdaCN: An Adaptive Cubic Newton Method for Nonconvex Stochastic Optimization

In this work, we introduce AdaCN, a novel adaptive cubic Newton method for nonconvex stochastic optimization. AdaCN dynamically captures the curvature of the loss landscape by diagonally approximated Hessian plus the norm of difference between previous two estimates. It only requires at most first o...

Descripción completa

Detalles Bibliográficos
Autores principales: Liu, Yan, Zhang, Maojun, Zhong, Zhiwei, Zeng, Xiangrong
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Hindawi 2021
Materias:
Research Article
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8598341/
https://www.ncbi.nlm.nih.gov/pubmed/34804146
http://dx.doi.org/10.1155/2021/5790608

Internet

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8598341/
https://www.ncbi.nlm.nih.gov/pubmed/34804146
http://dx.doi.org/10.1155/2021/5790608

Ejemplares similares