A Semi-Supervised Pipeline for Generalized Behavior Discovery from Animal-Borne Motion Time Series
Fatemeh Karimi Nejadasl, Judy Shamoun-Baranes, Eldar Rakhimberdiev · 2026
Read A Semi-Supervised Pipeline for Generalized Behavior Discovery from Animal-Borne Motion Time Series by Fatemeh Karimi Nejadasl, Judy Shamoun-Baranes, Eldar Rakhimberdiev (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-2602-02618.txt.
Description
Fatemeh Karimi Nejadasl and 2 co-authors hunt for entirely new animal behaviors hiding in unlabeled sensor data, in a 2026 arXiv paper. Gulls wearing motion sensors produce short multivariate snippets combining three-axis acceleration and GPS speed. Expert labels cover nine behavior categories, but labels are scarce, classes are imbalanced, and some behaviors may be missing from the list.
Their semi-supervised pipeline first learns an embedding from the labeled subset. Label-guided clustering then groups labeled and unlabeled snippets into candidate behavior clusters. A containment score decides whether a cluster is truly novel. The score uses density estimation with highest-density regions to measure how much a discovered cluster overlaps each known class. When a whole behavior was withheld from supervision, the method recovered it as a distinct cluster flagged by low overlap. A negative control with nothing novel produced consistently higher overlaps.
Catalog record
| CREATOR | Fatemeh Karimi Nejadasl, Judy Shamoun-Baranes, Eldar Rakhimberdiev |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-02618