X-VFL: A New Vertical Federated Learning Framework with Cross Completion and Decision Subspace Alignment

2025-08-07

AI Summary

Vertical federated learning (VFL) faces two key bottlenecks: misaligned data samples (i.e., feature incompleteness across clients) and the inability to support independent local inference at individual clients. To address these, we propose X-VFL—the first framework enabling both non-aligned vertical federated modeling and unilateral local inference. Our core contributions are: (1) the Cross-Completion (XCom) module, which collaboratively recovers missing features across clients; and (2) the Decision Subspace Alignment (DS-Align) mechanism, ensuring client-specific models operate within a shared decision subspace while retaining full autonomy for local inference. X-VFL integrates cross-client feature imputation, subspace alignment, and PAGE/SGD-based optimization, accompanied by theoretical convergence guarantees. Experiments on CIFAR-10 and MIMIC-III demonstrate accuracy improvements of 15% and 43%, respectively, significantly outperforming state-of-the-art VFL methods. These results validate X-VFL’s effectiveness and practicality in real-world heterogeneous healthcare and multi-source image scenarios.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Computer Vision: Learning & Optimization for CV
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing

Abstract

Vertical Federated Learning (VFL) enables collaborative learning by integrating disjoint feature subsets from multiple clients/parties. However, VFL typically faces two key challenges: i) the requirement for perfectly aligned data samples across all clients (missing features are not allowed); ii) the requirement for joint collaborative inference/prediction involving all clients (it does not support locally independent inference on a single client). To address these challenges, we propose X-VFL, a new VFL framework designed to deal with the non-aligned data samples with (partially) missing features and to support locally independent inference of new data samples for each client. In particular, we design two novel modules in X-VFL: Cross Completion (XCom) and Decision Subspace Alignment (DS-Align). XCom can complete/reconstruct missing features for non-aligned data samples by leveraging information from other clients. DS-Align aligns local features with completed and global features across all clients within the decision subspace, thus enabling locally independent inference at each client. Moreover, we provide convergence theorems for different algorithms used in training X-VFL, showing an $O(1/sqrt{T})$ convergence rate for SGD-type algorithms and an $O(1/T)$ rate for PAGE-type algorithms, where $T$ denotes the number of training update steps. Extensive experiments on real-world datasets demonstrate that X-VFL significantly outperforms existing methods, e.g., achieving a 15% improvement in accuracy on the image CIFAR-10 dataset and a 43% improvement on the medical MIMIC-III dataset. These results validate the practical effectiveness and superiority of X-VFL, particularly in scenarios involving partially missing features and locally independent inference.

Problem

• Addresses non-aligned data samples with missing features in VFL
• Enables locally independent inference for each client in VFL
• Improves accuracy in scenarios with partial missing features

Innovation

• Cross Completion reconstructs missing features collaboratively
• Decision Subspace Alignment enables local independent inference
• Convergence guarantees for SGD and PAGE training algorithms