Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.12451 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911614562729984 |
|---|---|
| author | Zheng, Vincent Zhihao Marcotte, Étienne Ashok, Arjun Williams, Andrew Robert Sun, Lijun Drouin, Alexandre Zantedeschi, Valentina |
| author_facet | Zheng, Vincent Zhihao Marcotte, Étienne Ashok, Arjun Williams, Andrew Robert Sun, Lijun Drouin, Alexandre Zantedeschi, Valentina |
| contents | Context-aided forecasting (CAF) holds promise for integrating domain knowledge and forward-looking information, enabling AI systems to surpass traditional statistical methods. However, recent empirical studies reveal a puzzling gap: multimodal models often fail to outperform their unimodal counterparts. We hypothesize that this underperformance stems from poor context quality in existing datasets, as verification is challenging. To address these limitations, we introduce a semi-synthetic data augmentation method that generates contexts both descriptive of temporal dynamics and verifiably complementary to numerical histories. This approach enables massive-scale dataset creation, resulting in CAF-7M, a corpus of 7 million context-augmented time series windows, including a rigorously verified test set. We demonstrate that semi-synthetic pre-training transfers effectively to real-world evaluation, and show clear evidence of context utilization. Our results suggest that dataset quality, rather than architectural limitations, has been the primary bottleneck in context-aided forecasting. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_12451 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Overcoming the Modality Gap in Context-Aided Forecasting Zheng, Vincent Zhihao Marcotte, Étienne Ashok, Arjun Williams, Andrew Robert Sun, Lijun Drouin, Alexandre Zantedeschi, Valentina Machine Learning Context-aided forecasting (CAF) holds promise for integrating domain knowledge and forward-looking information, enabling AI systems to surpass traditional statistical methods. However, recent empirical studies reveal a puzzling gap: multimodal models often fail to outperform their unimodal counterparts. We hypothesize that this underperformance stems from poor context quality in existing datasets, as verification is challenging. To address these limitations, we introduce a semi-synthetic data augmentation method that generates contexts both descriptive of temporal dynamics and verifiably complementary to numerical histories. This approach enables massive-scale dataset creation, resulting in CAF-7M, a corpus of 7 million context-augmented time series windows, including a rigorously verified test set. We demonstrate that semi-synthetic pre-training transfers effectively to real-world evaluation, and show clear evidence of context utilization. Our results suggest that dataset quality, rather than architectural limitations, has been the primary bottleneck in context-aided forecasting. |
| title | Overcoming the Modality Gap in Context-Aided Forecasting |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2603.12451 |