Capsule Routing via Variational Bayes

De Sousa Ribeiro, Fabio, Leontidis, Georgios and Kollias, Stefanos (2020) Capsule Routing via Variational Bayes. In: Thirty-fourth AAAI Conference on Artificial Intelligence, 7-12 February 2020, New York, USA.

Full content URL:

Capsule Routing via Variational Bayes
[img] PDF
AAAI-DeSousaRibeiroF.pdf - Whole Document

Item Type:Conference or Workshop contribution (Paper)
Item Status:Live Archive


Capsule networks are a recently proposed type of neural network shown to outperform alternatives in challenging shape recognition tasks. In capsule networks, scalar neurons are replaced with capsule vectors or matrices, whose entries represent different properties of objects. The relationships between objects and their parts are learned via trainable viewpoint-invariant transformation matrices, and the presence of a given object is decided by the level of agreement among votes from its parts. This interaction occurs between capsule layers and is a process called routing-by-agreement. In this paper, we propose a new capsule routing algorithm derived from Variational Bayes for fitting a mixture of transforming gaussians, and show it is possible transform our capsule network into a Capsule-VAE. Our Bayesian approach addresses some of the inherent weaknesses of MLE based models such as the variance-collapse by modelling uncertainty over capsule pose parameters. We outperform the state-of-the-art on smallNORB using 50% fewer capsules than previously reported, achieve competitive performances on CIFAR-10, Fashion-MNIST, SVHN, and demonstrate significant improvement in MNIST to affNIST generalisation over previous works.

Additional Information:Flagship/Top conference on AI/ML
Keywords:Deep Learning, Capsule Networks, Variational Inference, Variational Bayes
Subjects:G Mathematical and Computer Sciences > G400 Computer Science
G Mathematical and Computer Sciences > G760 Machine Learning
G Mathematical and Computer Sciences > G700 Artificial Intelligence
Divisions:College of Science > School of Computer Science
Related URLs:
ID Code:38864
Deposited On:03 Dec 2019 10:03

Repository Staff Only: item control page