Maximum-Volume Nonnegative Matrix Factorization PubID: pubid:
Olivier Vu Thanh, Nicolas Gillis · 2026
Read Maximum-Volume Nonnegative Matrix Factorization PubID: pubid: by Olivier Vu Thanh, Nicolas Gillis (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-2602-04795.txt.
Description
Olivier Vu Thanh and Nicolas Gillis flip a familiar factorization idea in a 2026 paper on nonnegative matrix factorization. The method splits a data matrix into two smaller nonnegative matrices, giving a readable summary. A popular variant shrinks the volume of one factor to force a unique answer.
Their maximum-volume version instead grows the volume of the other factor. It stays identifiable under the same noiseless terms but acts very differently with noise. In practice it pulls out sparser splits and avoids degenerate answers. The authors prove its largest-volume answers match clustering columns into disjoint groups. They give two algorithms and a normalized variant that did best. A hyperspectral unmixing case shows the approach at work.
Catalog record
| CREATOR | Olivier Vu Thanh, Nicolas Gillis |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 60 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-04795