Inference for Generative Capsule Models

TL;DR

This paper introduces a generative model for capsule networks, along with a variational inference algorithm, demonstrating superior performance over previous methods on geometric and face data.

Contribution

It formulates a generative capsule model and derives a variational inference algorithm, advancing the understanding and capability of capsule networks.

Findings

Outperforms stacked capsule autoencoders on constellation data

Successfully infers object transformations and part assignments

Demonstrates effectiveness on geometric and face datasets

Abstract

Capsule networks (see e.g. Hinton et al., 2018) aim to encode knowledge and reason about the relationship between an object and its parts. In this paper we specify a \emph{generative} model for such data, and derive a variational algorithm for inferring the transformation of each object and the assignments of observed parts to the objects. We apply this model to (i) data generated from multiple geometric objects like squares and triangles ("constellations"), and (ii) data from a parts-based model of faces. Recent work by Kosiorek et al. [2019] has used amortized inference via stacked capsule autoencoders (SCAEs) to tackle this problem -- our results show that we significantly outperform them where we can make comparisons (on the constellations data).