Federated learning is a decentralized machine learning approach where clients train models locally and share model updates to develop a global model. This enables low-resource devices to collaboratively build a high-quality model without requiring direct access to the raw training data. However, despite only sharing model updates, federated learning still faces several privacy vulnerabilities. One of the key threats is membership inference attacks, which target clients’ privacy by determining whether a specific example is part of the training set. These attacks can compromise sensitive information in real-world applications, such as medical diagnoses within a healthcare system. Although there has been extensive research on membership inference attacks, a comprehensive and up-to-date survey specifically focused on it within federated learning is still absent. To fill this gap, we categorize and summarize membership inference attacks and their corresponding defense strategies based on their characteristics in this setting. We introduce a unique taxonomy of existing attack research and provide a systematic overview of various countermeasures. For these studies, we thoroughly analyze the strengths and weaknesses of different approaches. Finally, we identify and discuss key future research directions for readers interested in advancing the field.

Query processing that preserves both the data privacy of the owner and the query privacy of the client is a new research problem. It shows increasing importance as cloud computing drives more businesses to outsource their data and querying services. However, most existing studies, including those on data outsourcing, address the data privacy and query privacy separately and cannot be applied to this problem. In this paper, we propose a holistic and efficient solution that comprises a secure traversal framework and an encryption scheme based on privacy homomorphism. The framework is scalable to large datasets by leveraging an index-based approach. Based on this framework, we devise secure protocols for processing typical queries such as k-nearest-neighbor queries (kNN) on R-tree index. Moreover, several optimization techniques are presented to improve the efficiency of the query processing protocols. Our solution is verified by both theoretical analysis and performance study.

Local differential privacy (LDP) is an emerging technique for privacy-preserving data collection without a trusted collector. Despite its strong privacy guarantee, LDP cannot be easily applied to real-world graph analysis tasks such as community detection and centrality analysis due to its high implementation complexity and low data utility. In this paper, we address these two issues by presenting LF-GDPR, the first LDP-enabled graph metric estimation framework for graph analysis. It collects two atomic graph metrics—the adjacency bit vector and node degree—from each node locally. LF-GDPR simplifies the job of implementing LDP-related steps (e.g., local perturbation, aggregation and calibration) for a graph metric estimation task by providing either a complete or a parameterized algorithm for each step. To address low data utility of LDP, it optimally allocates privacy budget between the two atomic metrics during data collection. To demonstrate the usage of LF-GDPR, we show use cases on two common graph analysis tasks, namely, clustering coefficient estimation and community detection. The privacy and utility achieved by LF-GDPR are verified through theoretical analysis and extensive experimental results.

The growing number of mobile and IoT devices has nourished many intelligent applications. In order to produce high-quality machine learning models, they constantly access and collect rich personal data such as photos, browsing history, and text messages. However, direct access to personal data has raised increasing public concerns about privacy risks and security breaches. To address these concerns, there are two emerging solutions to privacy-preserving machine learning, namely local differential privacy and federated machine learning. The former is a distributed data collection strategy where each client perturbs data locally before submitting to the server, whereas the latter is a distributed machine learning strategy to train models on mobile devices locally and merge their output (e.g., parameter updates of a model) through a control protocol. In this article, we conduct a comparative study on the efficiency and privacy of both solutions. Our results show that in a standard population and domain setting, both can achieve an optimal misclassification rate lower than 20% and federated machine learning generally performs better at the cost of higher client CPU usage. Nonetheless, local differential privacy can benefit more from a larger client population ($>$> 1k). As for privacy guarantee, local differential privacy also has flexible control over the data leakage.