Latent Space Score-based Diffusion Model for Probabilistic Multivariate Time Series Imputation
(2025) 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025- Abstract
Accurate imputation is essential for the reliability and success of downstream tasks. Recently, diffusion models have attracted great attention in this field. However, these models neglect the latent distribution in a lower-dimensional space derived from the observed data, which limits the generative capacity of the diffusion model. Additionally, dealing with the original missing data without labels becomes particularly problematic. To address these issues, we propose the Latent Space Score-Based Diffusion Model (LSSDM) for probabilistic multivariate time series imputation. Observed values are projected onto low-dimensional latent space and coarse values of the missing data are reconstructed without knowing their ground truth values by... (More)
Accurate imputation is essential for the reliability and success of downstream tasks. Recently, diffusion models have attracted great attention in this field. However, these models neglect the latent distribution in a lower-dimensional space derived from the observed data, which limits the generative capacity of the diffusion model. Additionally, dealing with the original missing data without labels becomes particularly problematic. To address these issues, we propose the Latent Space Score-Based Diffusion Model (LSSDM) for probabilistic multivariate time series imputation. Observed values are projected onto low-dimensional latent space and coarse values of the missing data are reconstructed without knowing their ground truth values by this unsupervised learning approach. Finally, the reconstructed values are fed into a conditional diffusion model to obtain the precise imputed values of the time series. In this way, LSSDM not only possesses the power to identify the latent distribution but also seamlessly integrates the diffusion model to obtain the high-fidelity imputed values and assess the uncertainty of the dataset. Experimental results demonstrate that LSSDM achieves superior imputation performance while also providing a better explanation and uncertainty analysis of the imputation mechanism. The website of the code is https://github.com/gorgen2020/LSSDMimputation.
(Less)
- author
- Liang, Guojun ; Abiri, Najmeh ; Hashemi, Atiye Sadat LU ; Lundström, Jens ; Byttner, Stefan and Tiwari, Prayag
- organization
- publishing date
- 2025
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- Diffusion model, imputation, multivariate time series, variational graph autoencoder
- host publication
- ICASSP 2025, 2025 IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
- editor
- Rao, Bhaskar D ; Trancoso, Isabel ; Sharma, Gaurav and Mehta, Neelesh B.
- publisher
- IEEE - Institute of Electrical and Electronics Engineers Inc.
- conference name
- 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025
- conference location
- Hyderabad, India
- conference dates
- 2025-04-06 - 2025-04-11
- external identifiers
-
- scopus:105003891405
- ISBN
- 9798350368741
- DOI
- 10.1109/ICASSP49660.2025.10888912
- language
- English
- LU publication?
- yes
- id
- ba804bc3-0bb9-4239-b6d2-54f9ef1e9c7f
- date added to LUP
- 2025-09-16 15:23:46
- date last changed
- 2025-10-14 11:36:57
@inproceedings{ba804bc3-0bb9-4239-b6d2-54f9ef1e9c7f,
abstract = {{<p>Accurate imputation is essential for the reliability and success of downstream tasks. Recently, diffusion models have attracted great attention in this field. However, these models neglect the latent distribution in a lower-dimensional space derived from the observed data, which limits the generative capacity of the diffusion model. Additionally, dealing with the original missing data without labels becomes particularly problematic. To address these issues, we propose the Latent Space Score-Based Diffusion Model (LSSDM) for probabilistic multivariate time series imputation. Observed values are projected onto low-dimensional latent space and coarse values of the missing data are reconstructed without knowing their ground truth values by this unsupervised learning approach. Finally, the reconstructed values are fed into a conditional diffusion model to obtain the precise imputed values of the time series. In this way, LSSDM not only possesses the power to identify the latent distribution but also seamlessly integrates the diffusion model to obtain the high-fidelity imputed values and assess the uncertainty of the dataset. Experimental results demonstrate that LSSDM achieves superior imputation performance while also providing a better explanation and uncertainty analysis of the imputation mechanism. The website of the code is https://github.com/gorgen2020/LSSDMimputation.</p>}},
author = {{Liang, Guojun and Abiri, Najmeh and Hashemi, Atiye Sadat and Lundström, Jens and Byttner, Stefan and Tiwari, Prayag}},
booktitle = {{ICASSP 2025, 2025 IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings}},
editor = {{Rao, Bhaskar D and Trancoso, Isabel and Sharma, Gaurav and Mehta, Neelesh B.}},
isbn = {{9798350368741}},
keywords = {{Diffusion model; imputation; multivariate time series; variational graph autoencoder}},
language = {{eng}},
publisher = {{IEEE - Institute of Electrical and Electronics Engineers Inc.}},
title = {{Latent Space Score-based Diffusion Model for Probabilistic Multivariate Time Series Imputation}},
url = {{http://dx.doi.org/10.1109/ICASSP49660.2025.10888912}},
doi = {{10.1109/ICASSP49660.2025.10888912}},
year = {{2025}},
}