Taming Normalizing Flows

TL;DR

This paper introduces a fast fine-tuning algorithm for Normalizing Flow models that allows precise control over generated image categories, enhancing privacy and debiasing without retraining from scratch.

Contribution

The authors present a novel taming method for Normalizing Flows enabling targeted output manipulation through quick fine-tuning, preserving quality and efficiency.

Findings

Effective in removing specific individuals from generated images

Maintains high image quality after taming process

Achieves desired output distribution with minimal fine-tuning time

Abstract

We propose an algorithm for taming Normalizing Flow models - changing the probability that the model will produce a specific image or image category. We focus on Normalizing Flows because they can calculate the exact generation probability likelihood for a given image. We demonstrate taming using models that generate human faces, a subdomain with many interesting privacy and bias considerations. Our method can be used in the context of privacy, e.g., removing a specific person from the output of a model, and also in the context of debiasing by forcing a model to output specific image categories according to a given target distribution. Taming is achieved with a fast fine-tuning process without retraining the model from scratch, achieving the goal in a matter of minutes. We evaluate our method qualitatively and quantitatively, showing that the generation quality remains intact, while the desired changes are applied.