Weighted Contrastive Learning for Anomaly-Aware Time-Series Forecasting
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_ | 1866909949593911296 |
|---|---|
| author | Ekstrand, Joel Mattsson, Tor Taghiyarrenani, Zahra Nowaczyk, Slawomir Lundström, Jens Lindén, Mikael |
| author_facet | Ekstrand, Joel Mattsson, Tor Taghiyarrenani, Zahra Nowaczyk, Slawomir Lundström, Jens Lindén, Mikael |
| contents | Reliable forecasting of multivariate time series under anomalous conditions is crucial in applications such as ATM cash logistics, where sudden demand shifts can disrupt operations. Modern deep forecasters achieve high accuracy on normal data but often fail when distribution shifts occur. We propose Weighted Contrastive Adaptation (WECA), a Weighted contrastive objective that aligns normal and anomaly-augmented representations, preserving anomaly-relevant information while maintaining consistency under benign variations. Evaluations on a nationwide ATM transaction dataset with domain-informed anomaly injection show that WECA improves SMAPE on anomaly-affected data by 6.1 percentage points compared to a normally trained baseline, with negligible degradation on normal data. These results demonstrate that WECA enhances forecasting reliability under anomalies without sacrificing performance during regular operations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_07569 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Weighted Contrastive Learning for Anomaly-Aware Time-Series Forecasting Ekstrand, Joel Mattsson, Tor Taghiyarrenani, Zahra Nowaczyk, Slawomir Lundström, Jens Lindén, Mikael Machine Learning Artificial Intelligence Reliable forecasting of multivariate time series under anomalous conditions is crucial in applications such as ATM cash logistics, where sudden demand shifts can disrupt operations. Modern deep forecasters achieve high accuracy on normal data but often fail when distribution shifts occur. We propose Weighted Contrastive Adaptation (WECA), a Weighted contrastive objective that aligns normal and anomaly-augmented representations, preserving anomaly-relevant information while maintaining consistency under benign variations. Evaluations on a nationwide ATM transaction dataset with domain-informed anomaly injection show that WECA improves SMAPE on anomaly-affected data by 6.1 percentage points compared to a normally trained baseline, with negligible degradation on normal data. These results demonstrate that WECA enhances forecasting reliability under anomalies without sacrificing performance during regular operations. |
| title | Weighted Contrastive Learning for Anomaly-Aware Time-Series Forecasting |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2512.07569 |