Unsupervised representation learning with recognition-parametrised probabilistic models

TL;DR

This paper introduces recognition-parametrised models (RPM), a flexible probabilistic framework for unsupervised learning that captures latent dependencies without explicit generative models, demonstrated on high-dimensional data tasks.

Contribution

The paper proposes RPM, a novel semi-parametric approach enabling exact maximum-likelihood learning for discrete latents and effective approximations for continuous latents.

Findings

Effective in high-dimensional image classification with weak supervision

Enables direct image-level latent Dirichlet allocation

Applies to multi-factorial spatiotemporal datasets with Gaussian process analysis

Abstract

We introduce a new approach to probabilistic unsupervised learning based on the recognition-parametrised model (RPM): a normalised semi-parametric hypothesis class for joint distributions over observed and latent variables. Under the key assumption that observations are conditionally independent given latents, the RPM combines parametric prior and observation-conditioned latent distributions with non-parametric observation marginals. This approach leads to a flexible learnt recognition model capturing latent dependence between observations, without the need for an explicit, parametric generative model. The RPM admits exact maximum-likelihood learning for discrete latents, even for powerful neural-network-based recognition. We develop effective approximations applicable in the continuous-latent case. Experiments demonstrate the effectiveness of the RPM on high-dimensional data, learning image classification from weak indirect supervision; direct image-level latent Dirichlet allocation; and recognition-parametrised Gaussian process factor analysis (RP-GPFA) applied to multi-factorial spatiotemporal datasets. The RPM provides a powerful framework to discover meaningful latent structure underlying observational data, a function critical to both animal and artificial intelligence.