English

Oracle Guided Image Synthesis with Relative Queries

Computer Vision and Pattern Recognition 2022-05-02 v1 Artificial Intelligence

Abstract

Isolating and controlling specific features in the outputs of generative models in a user-friendly way is a difficult and open-ended problem. We develop techniques that allow an oracle user to generate an image they are envisioning in their head by answering a sequence of relative queries of the form \textit{"do you prefer image aa or image bb?"} Our framework consists of a Conditional VAE that uses the collected relative queries to partition the latent space into preference-relevant features and non-preference-relevant features. We then use the user's responses to relative queries to determine the preference-relevant features that correspond to their envisioned output image. Additionally, we develop techniques for modeling the uncertainty in images' predicted preference-relevant features, allowing our framework to generalize to scenarios in which the relative query training set contains noise.

Keywords

Cite

@article{arxiv.2204.14189,
  title  = {Oracle Guided Image Synthesis with Relative Queries},
  author = {Alec Helbling and Christopher John Rozell and Matthew O'Shaughnessy and Kion Fallah},
  journal= {arXiv preprint arXiv:2204.14189},
  year   = {2022}
}

Comments

Published at the International Conference on Learning Representations 2022, Workshop on Deep Generative Models for Highly Structured Data

R2 v1 2026-06-24T11:02:48.770Z