Deep Latent Variable Model based Vertical Federated Learning with Flexible Alignment and Labeling Scenarios
Kihun Hong, Sejun Park, Ganguk Hwang · 2025
Read Deep Latent Variable Model based Vertical Federated Learning with Flexible Alignment and Labeling Scenarios by Kihun Hong, Sejun Park, Ganguk Hwang (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-2505-11035.txt.
Description
Kihun Hong, Sejun Park, and Ganguk Hwang present a 2025 framework for vertical federated learning under messy real-world conditions. In vertical setups, institutions hold different features for the same users, and learning must proceed without pooling private data. Existing methods assume few parties, perfectly aligned records, or labels everywhere. The authors recast alignment gaps as missing data and build one framework that trains and predicts under any alignment and labeling pattern.
Across 168 configurations, spanning four data sets and many missingness patterns at training and test time, their method beat every baseline in 160 cases. The average margin over the next-best rival was 9.6 percentage points. The authors claim the first framework to juggle arbitrary alignment, unlabeled data, and many parties at once.
Catalog record
| CREATOR | Kihun Hong, Sejun Park, Ganguk Hwang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 80 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-11035