A multi-view contrastive learning framework for spatial embeddings in risk modelling
Freek Holvoet, Christopher Blier-Wong, Katrien Antonio · 2025
Read A multi-view contrastive learning framework for spatial embeddings in risk modelling by Freek Holvoet, Christopher Blier-Wong, Katrien Antonio (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-2511-17954.txt.
Description
Freek Holvoet, Christopher Blier-Wong, and Katrien Antonio build spatial embeddings, compact numeric summaries of place, to improve insurance risk modeling. Climate, weather, and demographic data are rich but messy and high dimensional, so they are hard to feed into standard predictive models.
Their contrastive framework aligns satellite images and map features from across Europe with coordinate based codes, producing embeddings that any data set with coordinates can use. A French housing price case and a Belgian flood claims case both show better prediction and clearer risk classification than raw coordinates alone. The effects also generalize to regions without training data.
Catalog record
| CREATOR | Freek Holvoet, Christopher Blier-Wong, Katrien Antonio |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 67 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-17954