Training Data Attribution for Diffusion Models

TL;DR

This paper introduces a novel ensemble-based method to attribute and assess the influence of training data on the outputs of diffusion models, enhancing explainability and understanding of training data impact.

Contribution

The paper proposes a new ensemble approach that enables efficient attribution of training data influence on diffusion model outputs, addressing explainability challenges.

Findings

Ensembles can effectively identify influential training examples.

The approach allows for impact assessment of training data on generated samples.

Ensembles serve as valid generative models.

Abstract

Diffusion models have become increasingly popular for synthesizing high-quality samples based on training datasets. However, given the oftentimes enormous sizes of the training datasets, it is difficult to assess how training data impact the samples produced by a trained diffusion model. The difficulty of relating diffusion model inputs and outputs poses significant challenges to model explainability and training data attribution. Here we propose a novel solution that reveals how training data influence the output of diffusion models through the use of ensembles. In our approach individual models in an encoded ensemble are trained on carefully engineered splits of the overall training data to permit the identification of influential training examples. The resulting model ensembles enable efficient ablation of training data influence, allowing us to assess the impact of training data on model outputs. We demonstrate the viability of these ensembles as generative models and the validity of our approach to assessing influence.