The Necessity and Pitfall of Augmentation in Deep Learning: Observations During a Case Study in Triplet Learning for Coin Images

Daniel Soukup

Abstract

We conducted a case study on a subset of the MUSCLE CIS image benchmark of modern coins with the goal to assess the potential of deep embedding learning for generating representative CNN feature vectors of coin images, which are clustered class by class. In the course of training our models (CNN), we applied algorithmic rotational augmentation to the coin images to enforce rotational invariance. While augmentation is a usual procedure for regularizing deep learning models towards more geometric invariance, exactly that procedure revealed an interesting yet precarious pitfall in deep embedding learning: its susceptibility to interpolation errors. That interpolation bias results in distorted and ambiguous representation clusters of coin classes in the feature space, jeopardizing classification capabilities.

Download


Paper Citation


in Harvard Style

Soukup D. (2020). The Necessity and Pitfall of Augmentation in Deep Learning: Observations During a Case Study in Triplet Learning for Coin Images.In Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM, ISBN 978-989-758-397-1, pages 387-394. DOI: 10.5220/0008910303870394


in Bibtex Style

@conference{icpram20,
author={Daniel Soukup},
title={The Necessity and Pitfall of Augmentation in Deep Learning: Observations During a Case Study in Triplet Learning for Coin Images},
booktitle={Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,},
year={2020},
pages={387-394},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0008910303870394},
isbn={978-989-758-397-1},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,
TI - The Necessity and Pitfall of Augmentation in Deep Learning: Observations During a Case Study in Triplet Learning for Coin Images
SN - 978-989-758-397-1
AU - Soukup D.
PY - 2020
SP - 387
EP - 394
DO - 10.5220/0008910303870394