Deep Fast Embedded CapsNet: Going Faster with Deep-Caps

Abstract

Deep Capsule Network is a proven concept for understanding complex data in computer vision. Deep Capsule Networks achieved state-of-the-art accuracy Canadian institute for advanced research (CIFAR10), which is not achieved by shallow capsule networks. Despite all these accomplishments, Deep Capsule Networks are very slow due to the ‘Dynamic Routing’ algorithm in addition to their deep architecture. In this paper, the deep fast embedded capsule network (Deep-FECapsNet) is introduced. Deep-FECapsNet is a novel deep capsule network architecture that uses 1D convolution-based dynamic routing with a fast element-wise multiplication transformation process. It competes with state-of-the-art methods in terms of accuracy in the capsule domain and excels in terms of speed and reduced complexity. This is shown by the 58% reduction in trainable parameters and 64% decrease in the average epoch time in the training process. Experimental results show excellent and verified properties.