Radial basis function networks for convolutional neural networks to  learn similarity distance metric and improve interpretability

Amirian, Mohammadreza; Schwenker, Friedhelm

doi:10.1109/ACCESS.2020.3007337

Please use this identifier to cite or link to this item: https://doi.org/10.21256/zhaw-26284

Publication type:	Article in scientific journal
Type of review:	Peer review (publication)
Title:	Radial basis function networks for convolutional neural networks to learn similarity distance metric and improve interpretability
Authors:	Amirian, Mohammadreza Schwenker, Friedhelm
et. al:	No
DOI:	10.1109/ACCESS.2020.3007337 10.21256/zhaw-26284
Published in:	IEEE Access
Volume(Issue):	8
Page(s):	123087
Pages to:	123097
Issue Date:	2020
Publisher / Ed. Institution:	IEEE
ISSN:	2169-3536
Language:	English
Subjects:	Radial basis function neural network (RBF); Convolutional neural network (CNN); CNN-RBF; Supervised learning; Unsupervised learning; Similarity distance metric
Subject (DDC):	006: Special computer methods
Abstract:	Radial basis function neural networks (RBFs) are prime candidates for pattern classification and regression and have been used extensively in classical machine learning applications. However, RBFs have not been integrated into contemporary deep learning research and computer vision using conventional convolutional neural networks (CNNs) due to their lack of adaptability with modern architectures. In this paper, we adapt RBF networks as a classifier on top of CNNs by modifying the training process and introducing a new activation function to train modern vision architectures end-to-end for image classification. The specific architecture of RBFs enables the learning of a similarity distance metric to compare and find similar and dissimilar images. Furthermore, we demonstrate that using an RBF classifier on top of any CNN architecture provides new human-interpretable insights about the decision-making process of the models. Finally, we successfully apply RBFs to a range of CNN architectures and evaluate the results on benchmark computer vision datasets.
URI:	https://digitalcollection.zhaw.ch/handle/11475/26284
Fulltext version:	Published version
License (according to publishing contract):	CC BY 4.0: Attribution 4.0 International
Departement:	School of Engineering
Organisational Unit:	Institute of Applied Information Technology (InIT)
Appears in collections:	Publikationen School of Engineering

Files in This Item:

File	Description	Size	Format
2020_Amirian-Schwenker_Radial-basis-function-networks-learn-similarity-distance-metric-improve-interpretability.pdf		7.85 MB	Adobe PDF	View/Open

Show full item record

Amirian, M., & Schwenker, F. (2020). Radial basis function networks for convolutional neural networks to learn similarity distance metric and improve interpretability. IEEE Access, 8, 123087–123097. https://doi.org/10.1109/ACCESS.2020.3007337

Amirian, M. and Schwenker, F. (2020) ‘Radial basis function networks for convolutional neural networks to learn similarity distance metric and improve interpretability’, IEEE Access, 8, pp. 123087–123097. Available at: https://doi.org/10.1109/ACCESS.2020.3007337.

M. Amirian and F. Schwenker, “Radial basis function networks for convolutional neural networks to learn similarity distance metric and improve interpretability,” IEEE Access, vol. 8, pp. 123087–123097, 2020, doi: 10.1109/ACCESS.2020.3007337.

AMIRIAN, Mohammadreza und Friedhelm SCHWENKER, 2020. Radial basis function networks for convolutional neural networks to learn similarity distance metric and improve interpretability. IEEE Access. 2020. Bd. 8, S. 123087–123097. DOI 10.1109/ACCESS.2020.3007337

Amirian, Mohammadreza, and Friedhelm Schwenker. 2020. “Radial Basis Function Networks for Convolutional Neural Networks to Learn Similarity Distance Metric and Improve Interpretability.” IEEE Access 8: 123087–97. https://doi.org/10.1109/ACCESS.2020.3007337.

Amirian, Mohammadreza, and Friedhelm Schwenker. “Radial Basis Function Networks for Convolutional Neural Networks to Learn Similarity Distance Metric and Improve Interpretability.” IEEE Access, vol. 8, 2020, pp. 123087–97, https://doi.org/10.1109/ACCESS.2020.3007337.