Generalized Dilation Structures in Convolutional Neural Networks

Gavneet Chadha, Jan Reimann, Andreas Schwung

2021

Abstract

Convolutional neural networks are known to provide superior performance in various application fields such as image recognition, natural language processing and time series analysis owing to their strong ability to learn spatial and temporal features in the input domain. One of the most profound types of convolution kernels presented in literature is the dilated convolution kernel used primarily for aggregating information from a larger perspective or receptive field. However, the dilation rate and thereby the structure of the kernel has to be fixed a priori, which limits the flexibility of these convolution kernels. In this study, we propose a generalized dilation network where arbitrary dilation structures within a specific dilation rate can be learned. To this end, we derive an end-to-end learnable architecture for dilation layers using the constrained log-barrier method. We test the proposed architecture on various image recognition tasks by investigating and comparing with the SimpleNet architecture. The results illustrate the applicability of the generalized dilation layers and their superior performance.

Paper Citation

in Harvard Style

Chadha G., Reimann J. and Schwung A. (2021). Generalized Dilation Structures in Convolutional Neural Networks.In Proceedings of the 10th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM, ISBN 978-989-758-486-2, pages 79-88. DOI: 10.5220/0010302800790088

in Bibtex Style

@conference{icpram21,
author={Gavneet Chadha and Jan Reimann and Andreas Schwung},
title={Generalized Dilation Structures in Convolutional Neural Networks},
booktitle={Proceedings of the 10th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,},
year={2021},
pages={79-88},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010302800790088},
isbn={978-989-758-486-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 10th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,
TI - Generalized Dilation Structures in Convolutional Neural Networks
SN - 978-989-758-486-2
AU - Chadha G.
AU - Reimann J.
AU - Schwung A.
PY - 2021
SP - 79
EP - 88
DO - 10.5220/0010302800790088