Zero-Shot Fine-Grained Style Transfer: Leveraging Distributed Continuous Style Representations to Transfer To Unseen Styles
By Eric Michael Smith, Diana Gonzalez-Rico, Emily Dinan, Y-Lan Boureau. 2019
11 citations recorded by the source corpus.
Text style transfer is usually performed using attributes that can take a handful of discrete values (e.g., positive to negative reviews). In this work, we introduce an architecture that can leverage pre-trained consistent continuous distributed style representations and use them to transfer to an attribute unseen during training, without requiring any re-tuning of the style transfer model. We demonstrate the method by training an architecture to transfer text conveying one sentiment to another sentiment, using a fine-grained set of over 20 sentiment labels rather than the binary positive/negative often used in style transfer. Our experiments show that this model can then rewrite text to match a target sentiment that was unseen during training.
Subfields: Natural Language Processing
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Natural Language Processing
- Jason Aaron Edward Weston
- Y-Lan Boureau
- Kurt Shuster
- Eric Michael Smith
- Emily Dinan
- Arthur Szlam
- Da Ju
- Stephen Roller
Citations in the corpus
- Recipes for Safety in Open-domain Chatbots (2020, background): “Analysing the actual dialogue (Table 18) shows that control methods are capable of producing the desired style attributes, see also the work of Smith et al. (2019).”
- Open-Domain Conversational Agents: Current Progress, Open Problems, and Future Directions (2020, methodology): “Promising methods for controlled text generation (See et al., 2019) and text rewriting (Lample et al., 2019; Smith et al., 2019) could be refined to provide models more personally tailored to individual preferences, but are still not mature enough for that application.”
- Controlling Style in Generated Dialogue (2020, methodology): “The two elements of (1) noise, and (2) pairing of a noisy un-styled first guess with a target style to generate the desired response, are both present in our recent style transfer approaches based on added noise and back-translation (Lample et al., 2019; Smith et al., 2019).”