Handling dropout probability estimation in convolution neural networks using meta-heuristics

De Rosa, Gustavo H., Papa, João P. and Yang, Xin-She ORCID logoORCID: https://orcid.org/0000-0001-8231-5556 (2018) Handling dropout probability estimation in convolution neural networks using meta-heuristics. Soft Computing, 22 (18) . pp. 6147-6156. ISSN 1432-7643 [Article] (doi:10.1007/s00500-017-2678-4)

PDF - Final accepted version (with author's formatting)
Download (745kB) | Preview


Deep learning-based approaches have been paramount in recent years, mainly due to their outstanding results in several application domains, ranging from face and object recognition to handwritten digit identification. Convolutional Neural Networks (CNN) have attracted a considerable attention since they model the intrinsic and complex brain working mechanisms. However, one main shortcoming of such models concerns their overfitting problem, which prevents the network from predicting unseen data effectively. In this paper, we address this problem by means of properly selecting a regularization parameter known as Dropout in the context of CNNs using meta-heuristic-driven techniques. As far as we know, this is the first attempt to tackle this issue using this methodology. Additionally, we also take into account a default dropout parameter and a dropout-less CNN for comparison purposes. The results revealed that optimizing Dropout-based CNNs is worthwhile, mainly due to the easiness in finding suitable dropout probability values, without needing to set new parameters empirically.

Item Type: Article
Keywords (uncontrolled): Convolutional Neural Networks, Dropout, Meta-Heuristic Optimization
Research Areas: A. > School of Science and Technology > Design Engineering and Mathematics
Item ID: 22162
Notes on copyright: This is a post-peer-review, pre-copyedit version of an article published in Soft Computing. The final authenticated version is available online at Springer via http://dx.doi.org/10.1007/s00500-017-2678-4
Useful Links:
Depositing User: Xin-She Yang
Date Deposited: 26 Jun 2017 15:18
Last Modified: 29 Nov 2022 19:40
URI: https://eprints.mdx.ac.uk/id/eprint/22162

Actions (login required)

View Item View Item


Activity Overview
6 month trend
6 month trend

Additional statistics are available via IRStats2.