Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2404.06198 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909242180501504 |
|---|---|
| author | Ehrig, Claudia Sonnleitner, Benedikt Neumann, Ursula Cleophas, Catherine Forestier, Germain |
| author_facet | Ehrig, Claudia Sonnleitner, Benedikt Neumann, Ursula Cleophas, Catherine Forestier, Germain |
| contents | Pre-trained models have become pivotal in enhancing the efficiency and accuracy of time series forecasting on target data sets by leveraging transfer learning. While benchmarks validate the performance of model generalization on various target data sets, there is no structured research providing similarity and diversity measures to explain which characteristics of source and target data lead to transfer learning success. Our study pioneers in systematically evaluating the impact of source-target similarity and source diversity on zero-shot and fine-tuned forecasting outcomes in terms of accuracy, bias, and uncertainty estimation. We investigate these dynamics using pre-trained neural networks across five public source datasets, applied to forecasting five target data sets, including real-world wholesales data. We identify two feature-based similarity and diversity measures, finding that source-target similarity reduces forecasting bias, while source diversity improves forecasting accuracy and uncertainty estimation, but increases the bias. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_06198 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | The impact of data set similarity and diversity on transfer learning success in time series forecasting Ehrig, Claudia Sonnleitner, Benedikt Neumann, Ursula Cleophas, Catherine Forestier, Germain Machine Learning Pre-trained models have become pivotal in enhancing the efficiency and accuracy of time series forecasting on target data sets by leveraging transfer learning. While benchmarks validate the performance of model generalization on various target data sets, there is no structured research providing similarity and diversity measures to explain which characteristics of source and target data lead to transfer learning success. Our study pioneers in systematically evaluating the impact of source-target similarity and source diversity on zero-shot and fine-tuned forecasting outcomes in terms of accuracy, bias, and uncertainty estimation. We investigate these dynamics using pre-trained neural networks across five public source datasets, applied to forecasting five target data sets, including real-world wholesales data. We identify two feature-based similarity and diversity measures, finding that source-target similarity reduces forecasting bias, while source diversity improves forecasting accuracy and uncertainty estimation, but increases the bias. |
| title | The impact of data set similarity and diversity on transfer learning success in time series forecasting |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2404.06198 |