LGGNMLMay 22, 2018

Information Constraints on Auto-Encoding Variational Bayes

arXiv:1805.08672v4150 citations
Originality Incremental advance
AI Analysis

This work addresses the problem of learning invariant and interpretable representations for researchers in machine learning and computational biology, offering an incremental improvement by integrating kernel-based constraints into existing auto-encoding variational Bayes methods.

The paper tackles the challenge of imposing structural constraints like conditional independence in variational autoencoders by proposing a framework that uses kernel-based independence measures (dHSIC) to enforce independence between latent representations and nuisance factors. The method is applied to single-cell RNA sequencing data, where it outperforms state-of-the-art approaches.

Parameterizing the approximate posterior of a generative model with neural networks has become a common theme in recent machine learning research. While providing appealing flexibility, this approach makes it difficult to impose or assess structural constraints such as conditional independence. We propose a framework for learning representations that relies on Auto-Encoding Variational Bayes and whose search space is constrained via kernel-based measures of independence. In particular, our method employs the $d$-variable Hilbert-Schmidt Independence Criterion (dHSIC) to enforce independence between the latent representations and arbitrary nuisance factors. We show how to apply this method to a range of problems, including the problems of learning invariant representations and the learning of interpretable representations. We also present a full-fledged application to single-cell RNA sequencing (scRNA-seq). In this setting the biological signal is mixed in complex ways with sequencing errors and sampling effects. We show that our method out-performs the state-of-the-art in this domain.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes