VAFL: A Vertical Analytic Federated Learning Mechanism for Efficient Crowdsourced Services via Closed-Form Solutions
Organizations: Central South University, China · Beijing Normal University, China · Muroran Institute of Technology, Japan · University of Maryland, Baltimore County, USA
Abstract
Intelligent services often rely on complementary information held by different organizations, motivating Vertical Federated Learning without centralized raw-data pooling. We propose VAFL, a Vertical Analytic Federated Learning framework that fits classification and regression heads on fixed local representations through regularized least squares. An agent-assisted protected-computation workflow supports two-party and multi-party collaboration. To reduce fitting overhead, we develop a full-dimensional equivalent solve that applies block factors directly to the target matrix, avoiding construction of the full inverse. We further study two width-reduction strategies: local spectral projection with coefficient lifting, and full-dimensional local fitting with low-rank cross-party correction. The first compresses the prediction space, whereas the second retains local components outside the collaboration subspace. We establish their algebraic properties and analyze their computational scope. Experiments on five benchmarks show the accuracy--cost tradeoffs of these strategies and compatibility with diverse frozen backbones. Predictive performance and training time identify useful operating points and the limits of compression.
Figures & tables
| Method | Partition | Parties | Modalities | C & R | Non- iterative fitting | Cross- party products | Feature- block recursion | Frozen backbone evaluation | Projected head | Local fit + reduced coupling |
| Fed3R ( Fanì et al., 2024 ) | Samples | – | Images | ✗ | ✓ | ✗ | ✗ | ✓ | ✗ | ✗ |
| AFL ( He et al., 2025b ) | Samples | – | Images | ✗ | ✓ | ✗ | ✗ | ✓ | ✗ | ✗ |
| AFCL ( Tang et al., 2025b ) | Samples | – | Images | ✗ | ✓ | ✗ | ✗ | ✓ | ✗ | ✗ |
| CALM ( Fan et al., 2025 ) | Samples | – | Images | ✗ | ✓ | ✗ | ✗ | ✓ | ✗ | ✗ |
| PyVertical ( Romanini et al., 2021 ) | Features | Images | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | |
| SecureBoost ( Cheng et al., 2021 ) | Features | Tabular | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| Task | Dataset | Accuracy (%) | ||||||||
| Linear | XGB | RF | NN | VAFL (VAFL-E) | VAFL 256 | VAFL 128 | VAFL 64 | |||
| CLS | CIFAR-10 | |||||||||
| CIFAR-100 | ||||||||||
| Dataset | Frozen backbone | Linear | XGB | RF | NN | VAFL |
| CIFAR-10 | ResNet18 | |||||
| MobileNetV2 | ||||||
| ShuffleNetV2 ( ) | ||||||
| ViT-MAE-B | ||||||
| CIFAR-100 | ResNet18 | |||||
| MobileNetV2 |
Federated Learning
Federated learning trains shared models across data holders without requiring them to pool their raw data.
See all 1,037 papers