Denoising Likelihood Score Matching for Conditional Score-based Data Generation

TL;DR

This paper introduces Denoising Likelihood Score Matching, a new training objective for conditional score-based data generation that reduces score mismatch and improves sample quality on benchmark datasets.

Contribution

The paper proposes DLSM, a novel training method that better aligns estimated scores with true likelihood gradients, addressing a key mismatch issue in prior methods.

Findings

Outperforms previous methods on CIFAR-10 and CIFAR-100 benchmarks.

Reduces score mismatch, leading to higher quality generated samples.

Enhances the accuracy of conditional score modeling.

Abstract

Many existing conditional score-based data generation methods utilize Bayes' theorem to decompose the gradients of a log posterior density into a mixture of scores. These methods facilitate the training procedure of conditional score models, as a mixture of scores can be separately estimated using a score model and a classifier. However, our analysis indicates that the training objectives for the classifier in these methods may lead to a serious score mismatch issue, which corresponds to the situation that the estimated scores deviate from the true ones. Such an issue causes the samples to be misled by the deviated scores during the diffusion process, resulting in a degraded sampling quality. To resolve it, we formulate a novel training objective, called Denoising Likelihood Score Matching (DLSM) loss, for the classifier to match the gradients of the true log likelihood density. Our experimental evidence shows that the proposed method outperforms the previous methods on both Cifar-10 and Cifar-100 benchmarks noticeably in terms of several key evaluation metrics. We thus conclude that, by adopting DLSM, the conditional scores can be accurately modeled, and the effect of the score mismatch issue is alleviated.