How to Accelerate Capsule Convolutions in Capsule Networks

TL;DR

This paper introduces two CUDA-based acceleration schemes for capsule convolutions in CapsNets, significantly improving their computational efficiency by approximately four times.

Contribution

The paper proposes novel CUDA acceleration methods for capsule convolutions, addressing their incompatibility with existing deep learning frameworks and enhancing performance.

Findings

Achieved 4X speedup in capsule convolution computations

Demonstrated effectiveness on a custom CapsNet model

Provided a practical solution for integrating capsule convolutions into deep learning workflows

Abstract

How to improve the efficiency of routing procedures in CapsNets has been studied a lot. However, the efficiency of capsule convolutions has largely been neglected. Capsule convolution, which uses capsules rather than neurons as the basic computation unit, makes it incompatible with current deep learning frameworks' optimization solution. As a result, capsule convolutions are usually very slow with these frameworks. We observe that capsule convolutions can be considered as the operations of `multiplication of multiple small matrics' plus tensor-based combination. Based on this observation, we develop two acceleration schemes with CUDA APIs and test them on a custom CapsNet. The result shows that our solution achieves a 4X acceleration.