Latent Target Score Matching, with an application to Simulation-Based Inference
This addresses a specific bottleneck in simulation-based inference for researchers, though it appears incremental as an extension of existing methods.
The paper tackled the problem of high variance in denoising score matching for diffusion models at low noise levels by proposing Latent Target Score Matching, which leverages joint scores when clean data scores are inaccessible, resulting in improved variance, score accuracy, and sample quality across simulation-based inference tasks.
Denoising score matching (DSM) for training diffusion models may suffer from high variance at low noise levels. Target Score Matching (TSM) mitigates this when clean data scores are available, providing a low-variance objective. In many applications clean scores are inaccessible due to the presence of latent variables, leaving only joint signals exposed. We propose Latent Target Score Matching (LTSM), an extension of TSM to leverage joint scores for low-variance supervision of the marginal score. While LTSM is effective at low noise levels, a mixture with DSM ensures robustness across noise scales. Across simulation-based inference tasks, LTSM consistently improves variance, score accuracy, and sample quality.