Guided Learning of Nonconvex Models through Successive Functional Gradient Optimization

Rie Johnson; Tong Zhang

Guided Learning of Nonconvex Models through Successive Functional Gradient Optimization

Rie Johnson^*, Tong Zhang^*

^*Corresponding author for this work

Department of Mathematics

Research output: Contribution to conference › Conference Paper › peer-review

Abstract

This paper presents a framework of successive functional gradient optimization for training nonconvex models such as neural networks, where training is driven by mirror descent in a function space. We provide a theoretical analysis and empirical study of the training method derived from this framework. It is shown that the method leads to better performance than that of standard training techniques.

Original language	English
Pages	4921-4930
Publication status	Published - Jul 2020
Event	Proceedings of Machine Learning Research - Duration: 1 Jul 2020 → 1 Jul 2020

Conference

Conference	Proceedings of Machine Learning Research
Period	1/07/20 → 1/07/20

Access to Document

https://hdl.handle.net/1783.1/114129

Cite this

@conference{de4ed35a24be4a4c8244fe578d033ae7,

title = "Guided Learning of Nonconvex Models through Successive Functional Gradient Optimization",

abstract = "This paper presents a framework of successive functional gradient optimization for training nonconvex models such as neural networks, where training is driven by mirror descent in a function space. We provide a theoretical analysis and empirical study of the training method derived from this framework. It is shown that the method leads to better performance than that of standard training techniques.",

author = "Rie Johnson and Tong Zhang",

year = "2020",

month = jul,

language = "English",

pages = "4921--4930",

note = "Proceedings of Machine Learning Research ; Conference date: 01-07-2020 Through 01-07-2020",

}

TY - CONF

T1 - Guided Learning of Nonconvex Models through Successive Functional Gradient Optimization

AU - Johnson, Rie

AU - Zhang, Tong

PY - 2020/7

Y1 - 2020/7

N2 - This paper presents a framework of successive functional gradient optimization for training nonconvex models such as neural networks, where training is driven by mirror descent in a function space. We provide a theoretical analysis and empirical study of the training method derived from this framework. It is shown that the method leads to better performance than that of standard training techniques.

AB - This paper presents a framework of successive functional gradient optimization for training nonconvex models such as neural networks, where training is driven by mirror descent in a function space. We provide a theoretical analysis and empirical study of the training method derived from this framework. It is shown that the method leads to better performance than that of standard training techniques.

UR - https://www.webofscience.com/wos/woscc/full-record/WOS:000683178505005

M3 - Conference Paper

SP - 4921

EP - 4930

T2 - Proceedings of Machine Learning Research

Y2 - 1 July 2020 through 1 July 2020

ER -

Guided Learning of Nonconvex Models through Successive Functional Gradient Optimization

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Cite this