SE-DAE: Style-Enhanced Denoising Auto-Encoder for Unsupervised Text Style Transfer

TL;DR

SE-DAE is a novel style-enhanced denoising auto-encoder designed for unsupervised text style transfer, improving transfer quality without complex networks by using a new data refinement and style denoising mechanism.

Contribution

The paper introduces SE-DAE, a simple yet effective model that leverages high-quality pseudo-parallel data and novel style denoising to outperform existing methods in text style transfer.

Findings

Outperforms previous state-of-the-art models on benchmark datasets.

Achieves higher quality style transfer with less complexity.

Both automatic and human evaluations confirm its effectiveness.

Abstract

Text style transfer aims to change the style of sentences while preserving the semantic meanings. Due to the lack of parallel data, the Denoising Auto-Encoder (DAE) is widely used in this task to model distributions of different sentence styles. However, because of the conflict between the target of the conventional denoising procedure and the target of style transfer task, the vanilla DAE can not produce satisfying enough results. To improve the transferability of the model, most of the existing works combine DAE with various complicated unsupervised networks, which makes the whole system become over-complex. In this work, we design a novel DAE model named Style-Enhanced DAE (SE-DAE), which is specifically designed for the text style transfer task. Compared with previous complicated style-transfer models, our model do not consist of any complicated unsupervised networks, but only relies on the high-quality pseudo-parallel data generated by a novel data refinement mechanism. Moreover, to alleviate the conflict between the targets of the conventional denoising procedure and the style transfer task, we propose another novel style denoising mechanism, which is more compatible with the target of the style transfer task. We validate the effectiveness of our model on two style benchmark datasets. Both automatic evaluation and human evaluation show that our proposed model is highly competitive compared with previous strong the state of the art (SOTA) approaches and greatly outperforms the vanilla DAE.