AdaFed-LDR: Adaptive Federated Learning with Layerwise Dynamics Regularization for Robust Wi-Fi Localization
Kaito Harada, Hirofumi Natori, Makoto Koike, Hiroshi Mineno · 2026
Read AdaFed-LDR: Adaptive Federated Learning with Layerwise Dynamics Regularization for Robust Wi-Fi Localization by Kaito Harada, Hirofumi Natori, Makoto Koike, Hiroshi Mineno (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/pmc-13210543.txt.
Description
Kaito Harada and three co-authors present AdaFed-LDR, a learning system that locates devices indoors by Wi-Fi. Their 2026 paper uses signal fingerprints to pinpoint where a device sits. Pooling such data in one place raises privacy worries, so the learning is split across sites. Yet each site's signals differ, which makes shared learning unstable.
The system weights each site's input by how much its features drift, on the server side. On the client side, it holds early layers steady while letting deep layers adapt to local signals. Across eight indoor sites, error in known sites dropped sharply versus a standard method. In new sites the system also beat a rival method. The design balances steady accuracy at home sites with flexibility in new ones.
Catalog record
| CREATOR | Kaito Harada, Hirofumi Natori, Makoto Koike, Hiroshi Mineno |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 83 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |