User: Guest  Login
Title:

Demystification of Flat Minima and Generalisability of Deep Neural Networks

Document type:
Konferenzbeitrag
Contribution type:
Poster
Author(s):
Shen, Hao; Gottwald, Martin
Abstract:
Among many unsolved puzzles in theories of Deep Neural Networks (DNNs), generalisability is arguably one of the most puzzling mysteries of DNNs. In this work, we investigates the concept of sharpness/flatness of local minima of the error function, and its relationship to generalisability of DNNs. By defining the sharpness of local minima as the largest Eigenvalue of the Hessian, we identify four influencing factors contributing to the sharpness, while three factors are also found for controlli...     »
Dewey Decimal Classification:
620 Ingenieurwissenschaften
Book / Congress title:
International Conference on Machine Learning
Congress (additional information):
Understanding and Improving Generalization in Deep Learning
Date of congress:
14 Juni 2019
Year:
2019
Quarter:
2. Quartal
Year / month:
2019-06
Month:
Jun
Reviewed:
ja
Language:
en
WWW:
ICML 2019 Workshop Paper 67
TUM Institution:
Lehrstuhl für Datenverarbeitung
Last change:
02.08.2022
 BibTeX