Investigation of Capsule Networks Regarding their Potential of Explainability and Image Rankings

Felizia Quetscher, Christof Kaufmann, Jörg Frochte

2022

Abstract

Explainable Artificial Intelligence (AI) is a long-ranged goal, which can be approached from different viewpoints. One way is to simplify the complex AI model into an explainable one, another way uses post- processing to highlight the most important input features for the classification. In this work, we focus on the explanation of image classification using capsule networks with dynamic routing. We train a capsule network on the EMNIST letter dataset and examine the model regarding its explanatory potential. We show that the length of the class specific vectors (squash vectors) of the capsule network can be interpreted as predicted probability and it correlates with the agreement between the decoded image and the original image. We use the predicted probabilities to rank images within one class. By decoding different squash vectors, we visualize the interpretation of the image as the corresponding classes. Eventually, we create a set of modified letters to examine which features contribute to the perception of letters. We conclude that this decoding of squash vectors provides a quantifiable tool towards explainability in AI applications. The explanations are trustworthy through the relation between the capsule network’s prediction and the corresponding visualization.

Paper Citation

in Harvard Style

Quetscher F., Kaufmann C. and Frochte J. (2022). Investigation of Capsule Networks Regarding their Potential of Explainability and Image Rankings. In Proceedings of the 14th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART, ISBN 978-989-758-547-0, pages 343-351. DOI: 10.5220/0010821600003116

in Bibtex Style

@conference{icaart22,
author={Felizia Quetscher and Christof Kaufmann and Jörg Frochte},
title={Investigation of Capsule Networks Regarding their Potential of Explainability and Image Rankings},
booktitle={Proceedings of the 14th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART,},
year={2022},
pages={343-351},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010821600003116},
isbn={978-989-758-547-0},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 14th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART,
TI - Investigation of Capsule Networks Regarding their Potential of Explainability and Image Rankings
SN - 978-989-758-547-0
AU - Quetscher F.
AU - Kaufmann C.
AU - Frochte J.
PY - 2022
SP - 343
EP - 351
DO - 10.5220/0010821600003116