Beauty Learning and Counterfactual Inference

TL;DR

This paper presents a novel causal discovery framework using photo-realistic image editing and user experiments, exemplified by the beauty learning problem, to identify causal factors in complex systems.

Contribution

It introduces a new approach combining image editing and user studies for causal inference, applied to the previously metaphysical concept of beauty.

Findings

Causal effects from facial semantics to beauty were inferred.

Results align with existing empirical studies.

Framework has potential for broader causal inference applications.

Abstract

This work showcases a new approach for causal discovery by leveraging user experiments and recent advances in photo-realistic image editing, demonstrating a potential of identifying causal factors and understanding complex systems counterfactually. We introduce the beauty learning problem as an example, which has been discussed metaphysically for centuries and been proved exists, is quantifiable, and can be learned by deep models in our recent paper, where we utilize a natural image generator coupled with user studies to infer causal effects from facial semantics to beauty outcomes, the results of which also align with existing empirical studies. We expect the proposed framework for a broader application in causal inference.