Table 1 Comparisons for related work with proposed scheme
From: IHVFL: a privacy-enhanced intention-hiding vertical federated learning framework for medical data
References | Description | Addressed challenges |
|---|---|---|
Hardy et al (2017) | This article firstly proposed the FL situation when the data is distributed vertically, and gave a solution based on a collaborator | Privacy protection |
Yang et al (2019b) | This article removed the trusted third-party role, proposed a parallel LR model, and improves the scalability of the VFL system | Scalability in VFL |
Chen et al (2021) | This article proposed a VFL model CAESAR to solve the problem of high-dimensional sparse data in the field of risk control | Sparse data in specific domains |
Chen et al (2022) | According to the complex and impractical problem of data interpretation and evaluation in VFL, this article proposed a explainable VFL framework to evaluate the importance of the features | Explainability and evaluability |
Fu et al (2022) | This article revealed the hidden privacy risk in VFL model training and proposed a novel label inference attack method | Label privacy and attacks |
Liu et al (2020b) | For the problem of unbalanced distribution of data,This article proposed an asymmetric VFL method to protect the ID privacy of weak parties | ID privacy |
Sun et al (2021) | For the intersection membership privacy across privacy-sensitive organizations, this article proposed a VFL framework, allowing each party to preserve private sensitive membership information | Intersection membership privacy |
This article | For the privacy protection requirements of hiding model training intention, this article proposed an Intention-Hiding VFL framework to achieve privacy enhancement in VFL | Intention privacy |