An interpretable unsupervised capsule network via comprehensive contrastive learning and two-stage training

Limited attention has been given to unsupervised capsule networks (CapsNets) with contrastive learning due to the challenge of harmoniously learning interpretable primary and high-level capsules. To address this issue, we focus on three aspects: loss function, routing algorithm, and training strateg...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Pattern recognition Jg. 158; S. 111059
Hauptverfasser: Zeng, Ru, Song, Yan, Zhong, Yanjiu
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier Ltd 01.02.2025
Schlagworte:
Comprehensive contrastive loss
Interpretability
Primary capsules
Routing algorithm
Unsupervised capsule network
Unsupervised capsule network
Routing algorithm
Comprehensive contrastive loss
Interpretability
Primary capsules
ISSN:0031-3203
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Limited attention has been given to unsupervised capsule networks (CapsNets) with contrastive learning due to the challenge of harmoniously learning interpretable primary and high-level capsules. To address this issue, we focus on three aspects: loss function, routing algorithm, and training strategy. First, we propose a comprehensive contrastive loss to ensure consistency in learning both high-level and primary capsules across different objects. Next, we introduce an agreement-based routing mechanism for the activation of high-level capsules. Finally, we present a two-stage training strategy to resolve conflicts between multiple losses. Ablation experiments show that these methods all improve model performance. Results from linear evaluation and semi-supervised learning demonstrate that our model outperforms other CapsNets and convolutional neural networks in learning high-level capsules. Additionally, visualizing capsules provides insights into the primary capsules, which remain consistent across images and align with human vision. •Propose unsupervised CapsNet based on contrastive learning.•Develop a comprehensive contrastive loss for interpretable capsules.•Implement a two-stage training strategy to resolve loss function conflicts.•Design an agreement routing algorithm considering prediction similarity and distribution.
ISSN:0031-3203
DOI:10.1016/j.patcog.2024.111059