Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2510.02470 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911199257427968 |
|---|---|
| author | Jha, Ashish Ahmadi-Asl, Salman |
| author_facet | Jha, Ashish Ahmadi-Asl, Salman |
| contents | Training modern neural networks on large datasets is computationally and energy intensive. We present SAGE, a streaming data-subset selection method that maintains a compact Frequent Directions (FD) sketch of gradient geometry in $O(\ell D)$ memory and prioritizes examples whose sketched gradients align with a consensus direction. The approach eliminates $N \times N$ pairwise similarities and explicit $N \times \ell$ gradient stores, yielding a simple two-pass, GPU-friendly pipeline. Leveraging FD's deterministic approximation guarantees, we analyze how agreement scoring preserves gradient energy within the principal sketched subspace. Across multiple benchmarks, SAGE trains with small kept-rate budgets while retaining competitive accuracy relative to full-data training and recent subset-selection baselines, and reduces end-to-end compute and peak memory. Overall, SAGE offers a practical, constant-memory alternative that complements pruning and model compression for efficient training. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_02470 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SAGE: Streaming Agreement-Driven Gradient Sketches for Representative Subset Selection Jha, Ashish Ahmadi-Asl, Salman Machine Learning Training modern neural networks on large datasets is computationally and energy intensive. We present SAGE, a streaming data-subset selection method that maintains a compact Frequent Directions (FD) sketch of gradient geometry in $O(\ell D)$ memory and prioritizes examples whose sketched gradients align with a consensus direction. The approach eliminates $N \times N$ pairwise similarities and explicit $N \times \ell$ gradient stores, yielding a simple two-pass, GPU-friendly pipeline. Leveraging FD's deterministic approximation guarantees, we analyze how agreement scoring preserves gradient energy within the principal sketched subspace. Across multiple benchmarks, SAGE trains with small kept-rate budgets while retaining competitive accuracy relative to full-data training and recent subset-selection baselines, and reduces end-to-end compute and peak memory. Overall, SAGE offers a practical, constant-memory alternative that complements pruning and model compression for efficient training. |
| title | SAGE: Streaming Agreement-Driven Gradient Sketches for Representative Subset Selection |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2510.02470 |