Active Learning Classification from a Signal Separation Perspective
Hrushikesh Mhaskar 1 2 , Ryan O’Dowd 2 and Efstratios Tsoukanis 2 · 2025
Read Active Learning Classification from a Signal Separation Perspective by Hrushikesh Mhaskar 1 2 , Ryan O’Dowd 2 and Efstratios Tsoukanis 2 (2025) 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-2502-16425.txt.
Description
Sorting items into classes is often framed as fitting a function from features to labels. Hrushikesh Mhaskar, Ryan O'Dowd, and Efstratios Tsoukanis offer a different framing in this 2025 paper, borrowed from the task of pulling apart mixed signals. Their method looks for the support of each class, the region where that class lives, even when classes overlap.
The team tests the method on two real sets of satellite image data, Salinas and Indian Pines. With very few labeled points used for training, their active learning method holds its own against leading rivals. The result suggests that the signal view can cut the cost of labeling, which is a real burden in fields where experts must mark each training case by hand.
Catalog record
| CREATOR | Hrushikesh Mhaskar 1 2 , Ryan O’Dowd 2 and Efstratios Tsoukanis 2 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 17 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-16425