Towards Mode Balancing of Generative Models via Diversity Weights

TL;DR

This paper introduces diversity weights, a training scheme that enhances generative model diversity by balancing modes in the training data, addressing the need for more varied outputs in creative applications.

Contribution

The paper proposes diversity weights, a novel training method that shifts from pure data distribution fitting to mode balancing, improving output diversity in generative models.

Findings

Controlled experiments show increased diversity in outputs

Method effectively balances modes in training data

Potential applications in creative and artistic domains

Abstract

Large data-driven image models are extensively used to support creative and artistic work. Under the currently predominant distribution-fitting paradigm, a dataset is treated as ground truth to be approximated as closely as possible. Yet, many creative applications demand a diverse range of output, and creators often strive to actively diverge from a given data distribution. We argue that an adjustment of modelling objectives, from pure mode coverage towards mode balancing, is necessary to accommodate the goal of higher output diversity. We present diversity weights, a training scheme that increases a model's output diversity by balancing the modes in the training dataset. First experiments in a controlled setting demonstrate the potential of our method. We discuss connections of our approach to diversity, equity, and inclusion in generative machine learning more generally, and computational creativity specifically. An implementation of our algorithm is available at https://github.com/sebastianberns/diversity-weights