Fine-tuning Deep Belief Networks using Harmony Search

Author(s)

Papa, Joao Paulo ^[1] ; Scheirer, Walter ^[2] ; Cox, David Daniel ^[2]

Affiliation

^[1] UNESP Univ Estadual Paulista, Dept Comp, Bauru - Brazil

^[2] Harvard Univ, Ctr Brain Sci, Cambridge, MA 02138 - USA

• Document type: Journal article
• Source: APPLIED SOFT COMPUTING; v. 46, p. 875-885, SEP 2016.
• Web of Science Citations: 32

Abstract

In this paper, we deal with the problem of Deep Belief Networks (DBNs) parameters fine-tuning by means of a fast meta-heuristic approach named Harmony Search (HS). Although such deep learning-based technique has been widely used in the last years, more detailed studies about how to set its parameters may not be observed in the literature. We have shown we can obtain more accurate results comparing HS against with several of its variants, a random search and two variants of the well-known Hyperopt library. The experimental results were carried out in two public datasets considering the task of binary image reconstruction, three DBN learning algorithms and three layers. (C) 2015 Elsevier B.V. All rights reserved. (AU)

• FAPESP's process: 13/20387-7 - Hyperparameter optimization in deep learning arquitectures
• Grantee: João Paulo Papa
• Support Opportunities: Scholarships abroad - Research
• FAPESP's process: 14/16250-9 - On the parameter optimization in machine learning techniques: advances and paradigms
• Grantee: João Paulo Papa
• Support Opportunities: Regular Research Grants