Theoretical Analysis of Image-to-Image Translation with Adversarial Learning

TL;DR

This paper provides a geometric theoretical analysis of image-to-image translation with adversarial learning, explaining empirical phenomena and proposing conditions to control model generalization.

Contribution

It offers a novel geometric reformulation of the model, interprets empirical observations, and extends generalization theory to guide model design and dataset choices.

Findings

Geometric interpretation of the model explains empirical phenomena.

Derived a condition for controlling generalization capability.

Provided practical suggestions for model and dataset design.

Abstract

Recently, a unified model for image-to-image translation tasks within adversarial learning framework has aroused widespread research interests in computer vision practitioners. Their reported empirical success however lacks solid theoretical interpretations for its inherent mechanism. In this paper, we reformulate their model from a brand-new geometrical perspective and have eventually reached a full interpretation on some interesting but unclear empirical phenomenons from their experiments. Furthermore, by extending the definition of generalization for generative adversarial nets to a broader sense, we have derived a condition to control the generalization capability of their model. According to our derived condition, several practical suggestions have also been proposed on model design and dataset construction as a guidance for further empirical researches.