Are Time-Indexed Foundation Models the Future of Time Series Imputation?
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866910016790855680 |
|---|---|
| author | Naour, Etienne Le Nabil, Tahar Petralia, Adrien Agoua, Ghislain |
| author_facet | Naour, Etienne Le Nabil, Tahar Petralia, Adrien Agoua, Ghislain |
| contents | Foundation models for time series imputation remain largely unexplored. Recently, two such models, TabPFN-TS and MoTM, have emerged. These models share a common philosophy that places them within the family of time-indexed foundation models. This paper presents the first large-scale empirical study of these models for zero-shot imputation, which enables missing value recovery without retraining across a wide range of scenarios. We conduct extensive univariate experiments across 33 out-of-domain datasets (approximately 1.3M imputation windows) and evaluate their ability to integrate covariates at inference time to improve accuracy without fine-tuning. Our results demonstrate that time-indexed foundation models are a powerful and practical step toward achieving general-purpose, zero-shot imputation for real-world time series. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_05980 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Are Time-Indexed Foundation Models the Future of Time Series Imputation? Naour, Etienne Le Nabil, Tahar Petralia, Adrien Agoua, Ghislain Machine Learning Foundation models for time series imputation remain largely unexplored. Recently, two such models, TabPFN-TS and MoTM, have emerged. These models share a common philosophy that places them within the family of time-indexed foundation models. This paper presents the first large-scale empirical study of these models for zero-shot imputation, which enables missing value recovery without retraining across a wide range of scenarios. We conduct extensive univariate experiments across 33 out-of-domain datasets (approximately 1.3M imputation windows) and evaluate their ability to integrate covariates at inference time to improve accuracy without fine-tuning. Our results demonstrate that time-indexed foundation models are a powerful and practical step toward achieving general-purpose, zero-shot imputation for real-world time series. |
| title | Are Time-Indexed Foundation Models the Future of Time Series Imputation? |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2511.05980 |