# Topographic VAEs learn Equivariant Capsules

Source: [Topographic VAEs learn Equivariant Capsules](<https://www.sota2.com/research/paper/68fa6b991e209131cea2f325>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-25T23:52:06.013Z

Authors: T. Anderson Keller, Max Welling

Published: 2021-09-03T09:25:57.000Z

## Abstract

In this work we seek to bridge the concepts of topographic organization and equivariance in neural networks. To accomplish this, we introduce the Topographic VAE: a novel method for efficiently training deep generative models with topographically organized latent variables. We show that such a model indeed learns to organize its activations according to salient characteristics such as digit class, width, and style on MNIST. Furthermore, through topographic organization over time (i.e. temporal coherence), we demonstrate how predefined latent space transformation operators can be encouraged for observed transformed input sequences -- a primitive form of unsupervised learned equivariance. We demonstrate that this model successfully learns sets of approximately equivariant features (i.e. "capsules") directly from sequences and achieves higher likelihood on correspondingly transforming test sequences. Equivariance is verified quantitatively by measuring the approximate commutativity of the inference network and the sequence transformations. Finally, we demonstrate approximate equivariance to complex transformations, expanding upon the capabilities of existing group equivariant neural networks.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Disentanglement on Shapes3D (10% train)](<https://www.sota2.com/research/sota/disentanglement-on-shapes3d-10-percent-train.md>): Disentanglement; Shapes3D (10% train); VP Score: 88.27.

- [Disentanglement on Shapes3D 1% (train)](<https://www.sota2.com/research/sota/disentanglement-on-shapes3d-1-percent-train.md>): Disentanglement; Shapes3D 1% (train); VP Score: 68.39.

- [Disentanglement on MNIST (10% train)](<https://www.sota2.com/research/sota/disentanglement-on-mnist-10-percent-train.md>): Disentanglement; MNIST (10% train); VP Score: 89.91.

- [Disentanglement on MNIST (1% train set)](<https://www.sota2.com/research/sota/disentanglement-on-mnist-1-percent-train-set.md>): Disentanglement; MNIST (1% train set); VP Score: 88.15.

- [Equivariance Error Measurement on Issac3D](<https://www.sota2.com/research/sota/equivariance-error-measurement-on-issac3d.md>): Equivariance Error Measurement; Issac3D; Robot X-move Error: 8441.65.

- [Unsupervised Disentanglement on Falcor3D (Physical Factors)](<https://www.sota2.com/research/sota/unsupervised-disentanglement-on-falcor3d-physical-factors.md>): Unsupervised Disentanglement; Falcor3D; Lighting Intensity: 11477.81.

- [Representation Learning on MNIST](<https://www.sota2.com/research/sota/representation-learning-on-mnist.md>): Representation Learning; MNIST; Scaling Equivariance Error: 505.19.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2109.01394>)

[PDF](<https://arxiv.org/pdf/2109.01394.pdf>)
