Complement Submodular Information Measures for Balanced and Robust Data Selection
Rishabh Iyer · 2026
Read Complement Submodular Information Measures for Balanced and Robust Data Selection by Rishabh Iyer (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-24779.txt.
Description
Rishabh Iyer introduces Complement Submodular Information for even data selection. Classical goals score only the selected subset, ignoring links to the rest. Yet train-test splitting, benchmark building, and robust picking depend on even structure across both sides. The framework quantifies shared structure between a subset and its complement, yielding Facility Location, Graph Cut, and LogDet variants. Under bounded curvature it shows near-monotonicity with near-optimal guarantees. Tests show wins on hidden-slice picks, preserving rare tail structure while suppressing noisy outliers.
Catalog record
| CREATOR | Rishabh Iyer |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 43548 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.24779