An Algorithm for Routing Capsules in All Domains

TL;DR

This paper introduces a new routing algorithm for capsule networks that improves accuracy and efficiency across vision and language tasks, demonstrating state-of-the-art results and broad applicability.

Contribution

It presents a novel general-purpose routing algorithm for capsule networks, applicable to multiple domains, with demonstrated state-of-the-art performance.

Findings

Achieved 99.1% accuracy on smallNORB with fewer parameters

Attained 58.5% and 95.6% accuracy on Stanford Sentiment Treebank tasks

Provided code and replication instructions for the proposed method

Abstract

Building on recent work on capsule networks, we propose a new, general-purpose form of "routing by agreement" that activates output capsules in a layer as a function of their net benefit to use and net cost to ignore input capsules from earlier layers. To illustrate the usefulness of our routing algorithm, we present two capsule networks that apply it in different domains: vision and language. The first network achieves new state-of-the-art accuracy of 99.1% on the smallNORB visual recognition task with fewer parameters and an order of magnitude less training than previous capsule models, and we find evidence that it learns to perform a form of "reverse graphics." The second network achieves new state-of-the-art accuracies on the root sentences of the Stanford Sentiment Treebank: 58.5% on fine-grained and 95.6% on binary labels with a single-task model that routes frozen embeddings from a pretrained transformer as capsules. In both domains, we train with the same regime. Code is available at https://github.com/glassroom/heinsen_routing along with replication instructions.