Privacy-preserving biometric authentication (PPBA) enables client authentication without revealing sensitive bio-metric data, addressing privacy and security concerns. Many studies have proposed efficient cryptographic solutions to this problem based on secure multi-party computation, typically assuming a semi-honest adversary model, where all parties follow the protocol but may try to learn additional information. However, this assumption often falls short in real-world scenarios, where adversaries may behave maliciously and actively deviate from the protocol. In this paper, we propose, implement, and evaluate FLAME, a Flexible and Lightweight biometric Authentication scheme designed for a Malicious Environment. By hybridizing lightweight secret-sharing-family primitives within two-party computation, FLAME carefully designs a line of supporting protocols that incorporate integrity checks with rationally extra overhead. Additionally, FLAME enables server-side authentication with various similarity metrics through a crossmetric-compatible design, enhancing flexibility and robustness without requiring any changes to the server-side process. A rigorous theoretical analysis validates the correctness, security, and efficiency of FLAME. Extensive experiments highlight FLAME's superior efficiency, with a communication reduction by 97.61x 110.13x and a speedup of 2.72x 2.82x (resp. 6.58x 8.51x) in a LAN (resp. WAN) environment, when compared to the state-of-the-art work.

Stochastic Natural Gradient Variational Inference (NGVI) is a widely used method for approximating posterior distribution in probabilistic models. Despite its empirical success and foundational role in variational inference, its theoretical underpinnings remain limited, particularly in the case of non-conjugate likelihoods. While NGVI has been shown to be a special instance of Stochastic Mirror Descent, and recent work has provided convergence guarantees using relative smoothness and strong convexity for conjugate models, these results do not extend to the non-conjugate setting, where the variational loss becomes non-convex and harder to analyze. In this work, we focus on mean-field parameterization and advance the theoretical understanding of NGVI in three key directions. First, we derive sufficient conditions under which the variational loss satisfies relative smoothness with respect to a suitable mirror map. Second, leveraging this structure, we propose a modified NGVI algorithm incorporating non-Euclidean projections and prove its global non-asymptotic convergence to a stationary point. Finally, under additional structural assumptions about the likelihood, we uncover hidden convexity properties of the variational loss and establish fast global convergence of NGVI to a global optimum. These results provide new insights into the geometry and convergence behavior of NGVI in challenging inference settings.

Community search on attributed graphs has gained significant attention in recent years for its ability to provide meaningful and personalized results. Given a query community, a similar community search aims to identify the communities that are similar in structural and attributed characteristics to the query community. As real-world networks continue to grow in complexity and size, outsourcing graph data and search tasks to cloud servers not only saves local storage space but also significantly enhances search efficiency. Nonetheless, this inevitably raises concerns about data privacy since cloud servers are not completely trustworthy. In this paper, we research on privacy-preserving similar community search on graphs. We propose a privacy-preserving closest similar community search scheme for attributed graphs that leverages cloud servers to enhance search efficiency while safeguarding the sensitive information in the graph. We consider packaging communities using center vertices to evaluate relationships across communities without accessing details within the communities. To achieve this, we design a centrality score function that integrates attribute contribution and closeness centrality to identify the center vertex of a community. To ensure the security of sensitive information in the attribute graph, we construct three secure indexes for the original graph utilizing diverse cryptographic primitives. By searching secure indexes, cloud servers can answer the closest similar community searches without possessing any sensitive information about the attribute graph. We employ Paillier homomorphic cryptosystem and related protocols to support efficient and secure evaluation of the distance and similarity between two communities on secure indexes. The security analysis confirms that the proposed scheme can be against adaptive chosen-query attacks so as to achieve CQA2-security and experimental results demonstrate the efficiency of the proposed scheme.

The minimum community search is used to identify a minimum dense community that includes a specified vertex in a large network. It has gained significant attention because of its various applications in social-network analysis, e-commerce transactions, biological network modeling, and other areas. Nevertheless, how to realize privacy-preserving minimum community search remains unexplored up to now. In this paper, we initiate the first research on privacy-preserving approximate minimum community search. We propose an effective scheme that allows cloud servers to identify the smallest possible community while safeguarding the private information of the network. To ensure the privacy of sensitive information in the network, we employ obfuscation technology and graph encryption technology to construct two secure indexes instead of the original graph. To strike a balance between safeguarding private information and maintaining search efficiency, our scheme incorporates Bloom filters into the index and implements a two-step strategy on the secure indexes to achieve privacy-preserving approximate minimum community searches. Furthermore, to secure the privacy of the search result, we carefully design an array comparison protocol based on the BGN cryptosystem. This protocol enables cloud servers to perform privacy-preserving heuristic searches from the initial community without exposing any details about the approximate minimum community. The security analysis confirms that our scheme achieves CQA2-security for two non-colluding cloud servers. The experimental results based on real social networks show that the proposed scheme can efficiently handle approximate minimum community searches on large networks.