Abstract:Unsupervised style transfer aims to change the style of an input sentence while preserving its original content without using parallel training data. In current dominant approaches, owing to the lack of fine-grained control on the influence from the target style,they are unable to yield desirable output sentences. In this paper, we propose a novel attentional sequence-to-sequence (Seq2seq) model that dynamically exploits the relevance of each output word to the target style for unsupervised style transfer. Specifically, we first pretrain a style classifier, where the relevance of each input word to the original style can be quantified via layer-wise relevance propagation. In a denoising auto-encoding manner, we train an attentional Seq2seq model to reconstruct input sentences and repredict word-level previously-quantified style relevance simultaneously. In this way, this model is endowed with the ability to automatically predict the style relevance of each output word. Then, we equip the decoder of this model with a neural style component to exploit the predicted wordlevel style relevance for better style transfer. Particularly, we fine-tune this model using a carefully-designed objective function involving style transfer, style relevance consistency, content preservation and fluency modeling loss terms. Experimental results show that our proposed model achieves state-of-the-art performance in terms of both transfer accuracy and content preservation.

Towards Unsupervised Text Multi-Style Transfer with Parameter-Sharing Scheme.

UATST: Towards Unpaired Arbitrary Text-Guided Style Transfer with Cross-Space Modulation

TeSTNeRF: Text-Driven 3D Style Transfer Via Cross-Modal Learning.

Style Transfer as Unsupervised Machine Translation

MSSRNet: Manipulating Sequential Style Representation for Unsupervised Text Style Transfer

Transductive Learning for Unsupervised Text Style Transfer

A Dual Reinforcement Learning Framework for Unsupervised Text Style Transfer

Evaluate and Improve the Quality of Neural Style Transfer.

Diversified Patch-based Style Transfer with Shifted Style Normalization

Semi-supervised Text Style Transfer: Cross Projection in Latent Space

ST$^2$: Small-data Text Style Transfer via Multi-task Meta-Learning

Unsupervised Text Style Transfer via LLMs and Attention Masking with Multi-way Interactions

Style Transfer in Text: Exploration and Evaluation

Text Style Transfer Via Learning Style Instance Supported Latent Space

Prefix-Tuning Based Unsupervised Text Style Transfer

NAST: A Non-Autoregressive Generator with Word Alignment for Unsupervised Text Style Transfer

Style Transformer: Unpaired Text Style Transfer Without Disentangled Latent Representation

Balancing Effect of Training Dataset Distribution of Multiple Styles for Multi-Style Text Transfer

Style Transfer with Multi-iteration Preference Optimization

Exploring Contextual Word-level Style Relevance for Unsupervised Style Transfer

StyleFlow: Disentangle Latent Representations via Normalizing Flow for Unsupervised Text Style Transfer