Co-Regularized Adversarial Learning for Multi-Domain Text Classification

TL;DR

This paper introduces a co-regularized adversarial learning framework for multi-domain text classification, addressing key limitations of existing methods by enhancing domain alignment and feature discriminability, resulting in improved performance.

Contribution

The paper proposes a novel co-regularized adversarial learning approach with dual shared spaces and consistency regularization for better multi-domain text classification.

Findings

Outperforms state-of-the-art on two benchmarks

Effectively improves domain alignment and feature discriminability

Enhances robustness of predictions on unlabeled data

Abstract

Multi-domain text classification (MDTC) aims to leverage all available resources from multiple domains to learn a predictive model that can generalize well on these domains. Recently, many MDTC methods adopt adversarial learning, shared-private paradigm, and entropy minimization to yield state-of-the-art results. However, these approaches face three issues: (1) Minimizing domain divergence can not fully guarantee the success of domain alignment; (2) Aligning marginal feature distributions can not fully guarantee the discriminability of the learned features; (3) Standard entropy minimization may make the predictions on unlabeled data over-confident, deteriorating the discriminability of the learned features. In order to address the above issues, we propose a co-regularized adversarial learning (CRAL) mechanism for MDTC. This approach constructs two diverse shared latent spaces, performs domain alignment in each of them, and punishes the disagreements of these two alignments with respect to the predictions on unlabeled data. Moreover, virtual adversarial training (VAT) with entropy minimization is incorporated to impose consistency regularization to the CRAL method. Experiments show that our model outperforms state-of-the-art methods on two MDTC benchmarks.