Piecewise Latent Variables for Neural Variational Text Processing

TL;DR

This paper introduces a flexible piecewise constant distribution for neural variational models, enabling better representation of complex, multi-modal latent factors in natural language processing tasks, leading to improved performance.

Contribution

It proposes a novel, highly flexible piecewise constant distribution for latent variables, overcoming limitations of traditional Gaussian priors in neural variational models.

Findings

Enhanced document modeling performance

Improved natural language generation quality

Representation of complex latent factors

Abstract

Advances in neural variational inference have facilitated the learning of powerful directed graphical models with continuous latent variables, such as variational autoencoders. The hope is that such models will learn to represent rich, multi-modal latent factors in real-world data, such as natural language text. However, current models often assume simplistic priors on the latent variables - such as the uni-modal Gaussian distribution - which are incapable of representing complex latent factors efficiently. To overcome this restriction, we propose the simple, but highly flexible, piecewise constant distribution. This distribution has the capacity to represent an exponential number of modes of a latent target distribution, while remaining mathematically tractable. Our results demonstrate that incorporating this new latent distribution into different models yields substantial improvements in natural language processing tasks such as document modeling and natural language generation for dialogue.