Capsule network with shortcut routing

TL;DR

This paper proposes a new 'shortcut routing' mechanism for capsule networks that enhances computational efficiency while maintaining high classification accuracy, demonstrated on multiple datasets with significant reduction in calculations.

Contribution

Introduces 'shortcut routing' and fuzzy/attention-based methods that improve efficiency in capsule networks without sacrificing performance.

Findings

Achieved over 99% accuracy on Mnist dataset.

Reduced calculations by up to 2.5 times compared to EM routing.

Maintained comparable classification performance across datasets.

Abstract

This study introduces "shortcut routing," a novel routing mechanism in capsule networks that addresses computational inefficiencies by directly activating global capsules from local capsules, eliminating intermediate layers. An attention-based approach with fuzzy coefficients is also explored for improved efficiency. Experimental results on Mnist, smallnorb, and affNist datasets show comparable classification performance, achieving accuracies of 99.52%, 93.91%, and 89.02% respectively. The proposed fuzzy-based and attention-based routing methods significantly reduce the number of calculations by 1.42 and 2.5 times compared to EM routing, highlighting their computational advantages in capsule networks. These findings contribute to the advancement of efficient and accurate hierarchical pattern representation models.