Adaptive Noise Scheduling for Diffusion-based Imputation of Time-series Single-cell Data
Abstract
Accurate imputation of missing values in single-cell data remains challenging due to high dimensionality and heterogeneous structures, which can not be effectively handled by most traditional imputation methods. Diffusion-based generative models have shown strong performance by learning data distributions through progressive noising and denoising processes. However, these approaches typically rely on fixed noise schedules, and the impact of different noise scheduling strategies on model performance remains insufficiently understood. Our prior work introduced a time-varying blue noise based diffusion model for imputing missing values in time-series data and demonstrated superior performance over many existing methods. Building on this, we systematically investigate a range of noise scheduling strategies and examine how they influence both the noise injection process and overall imputation performance. Experimental results on single-cell RNA sequencing data show that the model is largely insensitive to the ordering of different noise types during injection, and that different noise schedules have limited impact on reconstruction accuracy, thereby providing greater flexibility in the design of noise schedules during model training. These findings offer a promising direction for advancing generative modeling in high-dimensional data analysis.
Related
- Diffusion-Model-Guided Adaptive Sampling of Rare Conformational States in Adenylate Kinase
- Single Molecule Diffusion Studies Reveal Rapid Translations Across Target Membrane Surfaces for both p40Phox PX domain and Class III PI3K VPS34 Complex 1, the Autophagy Lipid Kinase
- AnDi-J: diffusion analysis of single-particle tracking data in Fiji
- Single-protein diffusion in micrometer-size vesicles is governed by viscosity, surface charge, and membrane interactions
- Single-Cell Multi-Omics Reveals Dynamic Chromatin Remodeling of Adaptive Immunity During Prime-Boost Vaccination
BibTeX
@article{bishop2026adaptive,
title = {{Adaptive Noise Scheduling for Diffusion-based Imputation of Time-series Single-cell Data}},
author = {Graham Bishop and Gavin Edwards and Haijun Gong and Tong Si},
journal = {Proceedings of the ... International Conference on Bioinformatics and Biomedical Technology},
year = {2026},
month = oct,
doi = {10.1145/3822521.3822557},
url = {https://doi.org/10.1145/3822521.3822557},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Adaptive Noise Scheduling for Diffusion-based Imputation of Time-series Single-cell Data
AU - Graham Bishop
AU - Gavin Edwards
AU - Haijun Gong
AU - Tong Si
PY - 2026
DA - 2026/10/16
JO - Proceedings of the ... International Conference on Bioinformatics and Biomedical Technology
DO - 10.1145/3822521.3822557
UR - https://doi.org/10.1145/3822521.3822557
AB - Accurate imputation of missing values in single-cell data remains challenging due to high dimensionality and heterogeneous structures, which can not be effectively handled by most traditional imputation methods. Diffusion-based generative models have shown strong performance by learning data distributions through progressive noising and denoising processes. However, these approaches typically rely on fixed noise schedules, and the impact of different noise scheduling strategies on model performance remains insufficiently understood. Our prior work introduced a time-varying blue noise based diffusion model for imputing missing values in time-series data and demonstrated superior performance over many existing methods. Building on this, we systematically investigate a range of noise scheduling strategies and examine how they influence both the noise injection process and overall imputation performance. Experimental results on single-cell RNA sequencing data show that the model is largely insensitive to the ordering of different noise types during injection, and that different noise schedules have limited impact on reconstruction accuracy, thereby providing greater flexibility in the design of noise schedules during model training. These findings offer a promising direction for advancing generative modeling in high-dimensional data analysis.
ER -