Boost-and-Skip: A Simple Guidance-Free Diffusion for Minority Generation

TL;DR

Boost-and-Skip is a guidance-free diffusion method that efficiently generates minority samples by simple modifications, outperforming guidance-based methods with less computation, supported by strong theoretical and empirical evidence.

Contribution

It introduces a novel guidance-free approach with minimal changes to standard diffusion models, effectively promoting minority feature emergence.

Findings

Achieves comparable or better minority sample generation than guidance-based methods.

Requires significantly less computational resources.

Supported by theoretical and empirical validation.

Abstract

Minority samples are underrepresented instances located in low-density regions of a data manifold, and are valuable in many generative AI applications, such as data augmentation, creative content generation, etc. Unfortunately, existing diffusion-based minority generators often rely on computationally expensive guidance dedicated for minority generation. To address this, here we present a simple yet powerful guidance-free approach called Boost-and-Skip for generating minority samples using diffusion models. The key advantage of our framework requires only two minimal changes to standard generative processes: (i) variance-boosted initialization and (ii) timestep skipping. We highlight that these seemingly-trivial modifications are supported by solid theoretical and empirical evidence, thereby effectively promoting emergence of underrepresented minority features. Our comprehensive experiments demonstrate that Boost-and-Skip greatly enhances the capability of generating minority samples, even rivaling guidance-based state-of-the-art approaches while requiring significantly fewer computations. Code is available at https://github.com/soobin-um/BnS.