Evaluating Architectures and Hyperparameters of Self-supervised Network Projections

Tim Cech, Daniel Atzberger, Willy Scheibel, Rico Richter, Jürgen Döllner

2023

Abstract

Self-Supervised Network Projections (SSNP) are dimensionality reduction algorithms that produce low-dimensional layouts from high-dimensional data. By combining an autoencoder architecture with neighborhood information from a clustering algorithm, SSNP intend to learn an embedding that generates visually separated clusters. In this work, we extend an approach that uses cluster information as pseudo-labels for SSNP by taking outlier information into account. Furthermore, we investigate the influence of different autoencoders on the quality of the generated two-dimensional layouts. We report on two experiments on the autoencoder’s architecture and hyperparameters, respectively, measuring nine metrics on eight labeled datasets from different domains, e.g., Natural Language Processing. The results indicate that the model’s architecture and the choice of hyperparameter values can influence the layout with statistical significance, but none achieves the best result over all metrics. In addition, we found out that using outlier information for the pseudo-labeling approach can maintain global properties of the two-dimensional layout while trading-off local properties.

Download


Paper Citation


in Harvard Style

Cech T., Atzberger D., Scheibel W., Richter R. and Döllner J. (2023). Evaluating Architectures and Hyperparameters of Self-supervised Network Projections. In Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 3: IVAPP; ISBN 978-989-758-634-7, SciTePress, pages 187-194. DOI: 10.5220/0011699700003417


in Bibtex Style

@conference{ivapp23,
author={Tim Cech and Daniel Atzberger and Willy Scheibel and Rico Richter and Jürgen Döllner},
title={Evaluating Architectures and Hyperparameters of Self-supervised Network Projections},
booktitle={Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 3: IVAPP},
year={2023},
pages={187-194},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011699700003417},
isbn={978-989-758-634-7},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 3: IVAPP
TI - Evaluating Architectures and Hyperparameters of Self-supervised Network Projections
SN - 978-989-758-634-7
AU - Cech T.
AU - Atzberger D.
AU - Scheibel W.
AU - Richter R.
AU - Döllner J.
PY - 2023
SP - 187
EP - 194
DO - 10.5220/0011699700003417
PB - SciTePress