DeepAI AI Chat
Log In Sign Up

Unsupervised Representation Disentanglement of Text: An Evaluation on Synthetic Datasets

by   Lan Zhang, et al.

To highlight the challenges of achieving representation disentanglement for text domain in an unsupervised setting, in this paper we select a representative set of successfully applied models from the image domain. We evaluate these models on 6 disentanglement metrics, as well as on downstream classification tasks and homotopy. To facilitate the evaluation, we propose two synthetic datasets with known generative factors. Our experiments highlight the existing gap in the text domain and illustrate that certain elements such as representation sparsity (as an inductive bias), or representation coupling with the decoder could impact disentanglement. To the best of our knowledge, our work is the first attempt on the intersection of unsupervised representation disentanglement and text, and provides the experimental framework and datasets for examining future developments in this direction.


page 6

page 7


Text-Aware Single Image Specular Highlight Removal

Removing undesirable specular highlight from a single input image is of ...

T5Score: Discriminative Fine-tuning of Generative Evaluation Metrics

Modern embedding-based metrics for evaluation of generated text generall...

Do DALL-E and Flamingo Understand Each Other?

A major goal of multimodal research is to improve machine understanding ...

SMILE: Sequence-to-Sequence Domain Adaption with Minimizing Latent Entropy for Text Image Recognition

Training recognition models with synthetic images have achieved remarkab...

Motif Mining and Unsupervised Representation Learning for BirdCLEF 2022

We build a classification model for the BirdCLEF 2022 challenge using un...

Learning Invariant Representation for Unsupervised Image Restoration

Recently, cross domain transfer has been applied for unsupervised image ...

Code Repositories