Spatial Broadcast Decoder: A simple architecture for learning disentangled representations in VAEs


We present a simple neural rendering architecture that helps variational autoencoders (VAEs) learn disentangled representations. Instead of the deconvolutional network typically used in the decoder of VAEs, we tile (broadcast) the latent vector across space, concatenate fixed X- and Y-“coordinate” channels, and apply a fully convolutional network with 1x1 stride. This provides an architectural prior for dissociating positional from non-positional features in the latent distribution of VAEs, yet without providing any explicit supervision to this effect. We show that this architecture, which we term the Spatial Broadcast decoder, improves disentangling, reconstruction accuracy, and generalization to held-out regions in data space. It provides a particularly dramatic benefit when applied to datasets with small objects. We also emphasize a method for visualizing learned latent spaces that helped us diagnose our models and may prove useful for others aiming to assess data representations. Finally, we show the Spatial Broadcast Decoder is complementary to state-of-the-art (SOTA) disentangling techniques and when incorporated improves their performance.

ICLR 2019 Workshop on Learning from Limited Labeled Data

This was actually a precursor in using coordinate conditioning for image decoders (e.g. as used in VITs)

Loic Matthey
Loic Matthey
Staff Research Scientist in Machine Learning

ex-Neuroscientist working on Artificial General Intelligence at Google DeepMind. Unsupervised learning, structured generative models, concepts and how to make AI actually generalize is what I do.