Semi-supervised Stochastic Multi-Domain Learning using Variational Inference

TL;DR

This paper introduces a semi-supervised multi-domain learning approach using variational inference with stochastic gating, effectively capturing domain signals and improving NLP model performance across heterogeneous datasets.

Contribution

It proposes a novel latent variable model with stochastic gating for multi-domain learning, handling both domain-supervised and semi-supervised scenarios.

Findings

Significant performance improvements over benchmark domain adaptation methods.

Effective handling of heterogenous and semi-supervised domain data.

Comparison of discrete versus continuous latent variables.

Abstract

Supervised models of NLP rely on large collections of text which closely resemble the intended testing setting. Unfortunately matching text is often not available in sufficient quantity, and moreover, within any domain of text, data is often highly heterogenous. In this paper we propose a method to distill the important domain signal as part of a multi-domain learning system, using a latent variable model in which parts of a neural model are stochastically gated based on the inferred domain. We compare the use of discrete versus continuous latent variables, operating in a domain-supervised or a domain semi-supervised setting, where the domain is known only for a subset of training inputs. We show that our model leads to substantial performance improvements over competitive benchmark domain adaptation methods, including methods using adversarial learning.