Disentangled Unsupervised Image Translation via Restricted Information Flow

TL;DR

This paper introduces a novel unsupervised image translation method that infers domain-specific attributes without architectural biases, using information flow constraints to improve attribute manipulation accuracy across diverse datasets.

Contribution

The proposed approach infers domain-specific attributes from data by constraining information flow, avoiding restrictive architectural assumptions common in prior methods.

Findings

Achieves high manipulation accuracy on synthetic and natural datasets.

Effectively infers domain-specific attributes without attribute annotations.

Outperforms existing methods in disentangled image translation tasks.

Abstract

Unsupervised image-to-image translation methods aim to map images from one domain into plausible examples from another domain while preserving structures shared across two domains. In the many-to-many setting, an additional guidance example from the target domain is used to determine domain-specific attributes of the generated image. In the absence of attribute annotations, methods have to infer which factors are specific to each domain from data during training. Many state-of-art methods hard-code the desired shared-vs-specific split into their architecture, severely restricting the scope of the problem. In this paper, we propose a new method that does not rely on such inductive architectural biases, and infers which attributes are domain-specific from data by constraining information flow through the network using translation honesty losses and a penalty on the capacity of domain-specific embedding. We show that the proposed method achieves consistently high manipulation accuracy across two synthetic and one natural dataset spanning a wide variety of domain-specific and shared attributes.