Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.10237 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914552433606656 |
|---|---|
| author | Cornacchia, Elisabetta Mikulincer, Dan Mossel, Elchanan |
| author_facet | Cornacchia, Elisabetta Mikulincer, Dan Mossel, Elchanan |
| contents | We study how temporal correlations in the data can make certain sparse learning problems efficiently learnable by gradient-based methods. Our focus is on Boolean k-juntas, a canonical sparse learning problem known to pose barriers for gradient-based methods under independent uniform samples. We show that this picture changes when the samples are generated by a lazy random walk on the hypercube. In this setting, the temporal dependencies can be exploited by a two-layer ReLU network trained using stylized-SGD with a temporal-difference loss, which compares target and predicted increments across consecutive samples. For every fixed k, the resulting sample complexity is essentially linear in the ambient dimension d. By contrast, we show that for large-batch gradient methods using standard convex pointwise losses, temporal correlations do not provide the same advantage. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_10237 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | The Benefits of Temporal Correlations: SGD Learns k-Juntas from Random Walks Efficiently Cornacchia, Elisabetta Mikulincer, Dan Mossel, Elchanan Machine Learning We study how temporal correlations in the data can make certain sparse learning problems efficiently learnable by gradient-based methods. Our focus is on Boolean k-juntas, a canonical sparse learning problem known to pose barriers for gradient-based methods under independent uniform samples. We show that this picture changes when the samples are generated by a lazy random walk on the hypercube. In this setting, the temporal dependencies can be exploited by a two-layer ReLU network trained using stylized-SGD with a temporal-difference loss, which compares target and predicted increments across consecutive samples. For every fixed k, the resulting sample complexity is essentially linear in the ambient dimension d. By contrast, we show that for large-batch gradient methods using standard convex pointwise losses, temporal correlations do not provide the same advantage. |
| title | The Benefits of Temporal Correlations: SGD Learns k-Juntas from Random Walks Efficiently |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2605.10237 |