Rui Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Rui Zhang has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
- Kui Ren
Zhejiang University
5 shared publications - Jian Liu
Zhejiang University
4 shared publications - Nandini Asokan
University of Waterloo
3 shared publications - Zhihua Tian
Zhejiang University, Hangzhou, China
3 shared publications Sebastian Szyller
Zhejiang University Zhejiang University Intel Labs &Aalto University Zhejiang University University of Waterloo Aalto University
2 shared publications- Xiaoyang Hou
School of Cyber Science and Technology, Zhejiang University, Hangzhou, China
2 shared publications
Latest Publications
- 22 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (3) , 2028-2039
Toward Federated Learning of Deep Graph Neural Networks
Federated graph learning (FGL) aims to collaboratively train graph neural networks (GNNs) among multiple clients, where each client owns a subgraph of a global model. A key challenge in FGL arises from the possible interconnections between nodes distributed across different subgraphs, leading to an incomplete capture of neighborhood knowledge within the graph. Existing FGL frameworks attempt to learn missing neighborhood knowledge by generating pseudo nodes or transmitting missing node embedding directly across clients, which is either only suitable to 1-hop neighbor nodes or comes with high communication costs when training deeper GNNs. In this paper, we propose a novel framework for FGL named$\text{Fed}^{2}\text{GNN}$ that could fully capture neighborhood knowledge while achieving low communication costs. More specifically, we propose ego-tree, a new graph structure that is easy to build and allows us to reconstruct the neighborhood faithfully. Furthermore, we design an encoder-decoder-based method to build ego-tree. The encoder enables clients to transmit encoded information essential for tree construction with minimal communication costs, while the decoder empowers clients to build the ego-tree by decoding the received information. Extensive experiments on real-world network datasets show the effectiveness of our framework for training deep GNNs and about 100× less communication compared to prior works.- 22 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (3) , 2028-2039
Toward Federated Learning of Deep Graph Neural Networks
Federated graph learning (FGL) aims to collaboratively train graph neural networks (GNNs) among multiple clients, where each client owns a subgraph of a global model. A key challenge in FGL arises from the possible interconnections between nodes distributed across different subgraphs, leading to an incomplete capture of neighborhood knowledge within the graph. Existing FGL frameworks attempt to learn missing neighborhood knowledge by generating pseudo nodes or transmitting missing node embedding directly across clients, which is either only suitable to 1-hop neighbor nodes or comes with high communication costs when training deeper GNNs. In this paper, we propose a novel framework for FGL named$\text{Fed}^{2}\text{GNN}$ that could fully capture neighborhood knowledge while achieving low communication costs. More specifically, we propose ego-tree, a new graph structure that is easy to build and allows us to reconstruct the neighborhood faithfully. Furthermore, we design an encoder-decoder-based method to build ego-tree. The encoder enables clients to transmit encoded information essential for tree construction with minimal communication costs, while the decoder empowers clients to build the ego-tree by decoding the received information. Extensive experiments on real-world network datasets show the effectiveness of our framework for training deep GNNs and about 100× less communication compared to prior works.- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 22 (3) , 3093-3104
Attributed Graph Clustering in Collaborative Settings
Graph clustering is an unsupervised machine learning method that partitions the nodes in a graph into different groups. Despite achieving significant progress in exploiting both attributed and structured data information, graph clustering methods often face practical challenges related to data isolation. Moreover, the absence of collaborative methods for graph clustering limits their effectiveness. In this paper, we propose a collaborative graph clustering framework for attributed graphs, supporting attributed graph clustering over vertically partitioned data with different participants holding distinct features of the same data. Our method leverages a novel technique that reduces the sample space, improving the efficiency of the attributed graph clustering method. Furthermore, we compare our method to its centralized counterpart under a proximity condition, demonstrating that the successful local results of each participant contribute to the overall success of the collaboration. We fully implement our approach and evaluate its utility and efficiency by conducting experiments on four public datasets. The results demonstrate that our method achieves comparable accuracy levels to centralized attributed graph clustering methods. Our collaborative graph clustering framework provides an efficient and effective solution for graph clustering challenges related to data isolation.
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 22 (3) , 3093-3104
Attributed Graph Clustering in Collaborative Settings
Graph clustering is an unsupervised machine learning method that partitions the nodes in a graph into different groups. Despite achieving significant progress in exploiting both attributed and structured data information, graph clustering methods often face practical challenges related to data isolation. Moreover, the absence of collaborative methods for graph clustering limits their effectiveness. In this paper, we propose a collaborative graph clustering framework for attributed graphs, supporting attributed graph clustering over vertically partitioned data with different participants holding distinct features of the same data. Our method leverages a novel technique that reduces the sample space, improving the efficiency of the attributed graph clustering method. Furthermore, we compare our method to its centralized counterpart under a proximity condition, demonstrating that the successful local results of each participant contribute to the overall success of the collaboration. We fully implement our approach and evaluate its utility and efficiency by conducting experiments on four public datasets. The results demonstrate that our method achieves comparable accuracy levels to centralized attributed graph clustering methods. Our collaborative graph clustering framework provides an efficient and effective solution for graph clustering challenges related to data isolation.
- 24 August 2024
- Published by Springer Nature in Lifetime Data Analysis
- Vol. 30 (4) , 742-775
Copula-based analysis of dependent current status data with semiparametric linear transformation model
This paper discusses regression analysis of current status data with dependent censoring, a problem that often occurs in many areas such as cross-sectional studies, epidemiological investigations and tumorigenicity experiments. Copula model-based methods are commonly employed to tackle this issue. However, these methods often face challenges in terms of model and parameter identification. The primary aim of this paper is to propose a copula-based analysis for dependent current status data, where the association parameter is left unspecified. Our method is based on a general class of semiparametric linear transformation models and parametric copulas. We demonstrate that the proposed semiparametric model is identifiable under certain regularity conditions from the distribution of the observed data. For inference, we develop a sieve maximum likelihood estimation method, using Bernstein polynomials to approximate the nonparametric functions involved. The asymptotic consistency and normality of the proposed estimators are established. Finally, to demonstrate the effectiveness and practical applicability of our method, we conduct an extensive simulation study and apply the proposed method to a real data example.
- 24 August 2024
- Published by Springer Nature in Lifetime Data Analysis
- Vol. 30 (4) , 742-775
Copula-based analysis of dependent current status data with semiparametric linear transformation model
This paper discusses regression analysis of current status data with dependent censoring, a problem that often occurs in many areas such as cross-sectional studies, epidemiological investigations and tumorigenicity experiments. Copula model-based methods are commonly employed to tackle this issue. However, these methods often face challenges in terms of model and parameter identification. The primary aim of this paper is to propose a copula-based analysis for dependent current status data, where the association parameter is left unspecified. Our method is based on a general class of semiparametric linear transformation models and parametric copulas. We demonstrate that the proposed semiparametric model is identifiable under certain regularity conditions from the distribution of the observed data. For inference, we develop a sieve maximum likelihood estimation method, using Bernstein polynomials to approximate the nonparametric functions involved. The asymptotic consistency and normality of the proposed estimators are established. Finally, to demonstrate the effectiveness and practical applicability of our method, we conduct an extensive simulation study and apply the proposed method to a real data example.
- Open Access
- 4 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 305-316
Espresso: Robust Concept Filtering in Text-to-Image Models
- 4 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 305-316
Open AccessEspresso: Robust Concept Filtering in Text-to-Image Models
- Open Access
False Claims against Model Ownership Resolution
Deep neural network (DNN) models are valuable intellectual property of model owners, constituting a competitive advantage. Therefore, it is crucial to develop techniques to protect against model theft. Model ownership resolution (MOR) is a class of techniques that can deter model theft. A MOR scheme enables an accuser to assert an ownership claim for a suspect model by presenting evidence, such as a watermark or fingerprint, to show that the suspect model was stolen or derived from a source model owned by the accuser. Most of the existing MOR schemes prioritize robustness against malicious suspects, ensuring that the accuser will win if the suspect model is indeed a stolen model. In this paper, we show that common MOR schemes in the literature are vulnerable to a different, equally important but insufficiently explored, robustness concern: a malicious accuser. We show how malicious accusers can successfully make false claims against independent suspect models that were not stolen. Our core idea is that a malicious accuser can deviate (without detection) from the specified MOR process by finding (transferable) adversarial examples that successfully serve as evidence against independent suspect models. To this end, we first generalize the procedures of common MOR schemes and show that, under this generalization, defending against false claims is as challenging as preventing (transferable) adversarial examples. Via systematic empirical evaluation, we show that our false claim attacks always succeed in the MOR schemes that follow our generalization, including in a real-world model: Amazon's Rekognition API.
Open AccessFalse Claims against Model Ownership Resolution
Deep neural network (DNN) models are valuable intellectual property of model owners, constituting a competitive advantage. Therefore, it is crucial to develop techniques to protect against model theft. Model ownership resolution (MOR) is a class of techniques that can deter model theft. A MOR scheme enables an accuser to assert an ownership claim for a suspect model by presenting evidence, such as a watermark or fingerprint, to show that the suspect model was stolen or derived from a source model owned by the accuser. Most of the existing MOR schemes prioritize robustness against malicious suspects, ensuring that the accuser will win if the suspect model is indeed a stolen model. In this paper, we show that common MOR schemes in the literature are vulnerable to a different, equally important but insufficiently explored, robustness concern: a malicious accuser. We show how malicious accusers can successfully make false claims against independent suspect models that were not stolen. Our core idea is that a malicious accuser can deviate (without detection) from the specified MOR process by finding (transferable) adversarial examples that successfully serve as evidence against independent suspect models. To this end, we first generalize the procedures of common MOR schemes and show that, under this generalization, defending against false claims is as challenging as preventing (transferable) adversarial examples. Via systematic empirical evaluation, we show that our false claim attacks always succeed in the MOR schemes that follow our generalization, including in a real-world model: Amazon's Rekognition API.
- Open Access
On the Robustness of Dataset Inference
Machine learning (ML) models are costly to train as they can require a significant amount of data, computational resources and technical expertise. Thus, they constitute valuable intellectual property that needs protection from adversaries wanting to steal them. Ownership verification techniques allow the victims of model stealing attacks to demonstrate that a suspect model was in fact stolen from theirs. Although a number of ownership verification techniques based on watermarking or fingerprinting have been proposed, most of them fall short either in terms of security guarantees (well-equipped adversaries can evade verification) or computational cost. A fingerprinting technique, Dataset Inference (DI), has been shown to offer better robustness and efficiency than prior methods. The authors of DI provided a correctness proof for linear (suspect) models. However, in a subspace of the same setting, we prove that DI suffers from high false positives (FPs) -- it can incorrectly identify an independent model trained with non-overlapping data from the same distribution as stolen. We further prove that DI also triggers FPs in realistic, non-linear suspect models. We then confirm empirically that DI in the black-box setting leads to FPs, with high confidence. Second, we show that DI also suffers from false negatives (FNs) -- an adversary can fool DI (at the cost of incurring some accuracy loss) by regularising a stolen model's decision boundaries using adversarial training, thereby leading to an FN. To this end, we demonstrate that black-box DI fails to identify a model adversarially trained from a stolen dataset -- the setting where DI is the hardest to evade. Finally, we discuss the implications of our findings, the viability of fingerprinting-based ownership verification in general, and suggest directions for future work.
Open AccessOn the Robustness of Dataset Inference
Machine learning (ML) models are costly to train as they can require a significant amount of data, computational resources and technical expertise. Thus, they constitute valuable intellectual property that needs protection from adversaries wanting to steal them. Ownership verification techniques allow the victims of model stealing attacks to demonstrate that a suspect model was in fact stolen from theirs. Although a number of ownership verification techniques based on watermarking or fingerprinting have been proposed, most of them fall short either in terms of security guarantees (well-equipped adversaries can evade verification) or computational cost. A fingerprinting technique, Dataset Inference (DI), has been shown to offer better robustness and efficiency than prior methods. The authors of DI provided a correctness proof for linear (suspect) models. However, in a subspace of the same setting, we prove that DI suffers from high false positives (FPs) -- it can incorrectly identify an independent model trained with non-overlapping data from the same distribution as stolen. We further prove that DI also triggers FPs in realistic, non-linear suspect models. We then confirm empirically that DI in the black-box setting leads to FPs, with high confidence. Second, we show that DI also suffers from false negatives (FNs) -- an adversary can fool DI (at the cost of incurring some accuracy loss) by regularising a stolen model's decision boundaries using adversarial training, thereby leading to an FN. To this end, we demonstrate that black-box DI fails to identify a model adversarially trained from a stolen dataset -- the setting where DI is the hardest to evade. Finally, we discuss the implications of our findings, the viability of fingerprinting-based ownership verification in general, and suggest directions for future work.
- 16 May 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 21 (3) , 1274-1285
${\sf FederBoost}$: Private Federated Learning for GBDT
Federated Learning (FL) has been an emerging trend in machine learning and artificial intelligence. It allows multiple participants to collaboratively train a better global model and offers a privacy-aware paradigm for model training since it does not require participants to release their original training data. However, existing FL solutions for vertically partitioned data or decision trees require heavy cryptographic operations. In this paper, we propose a framework named $\textsf {FederBoost}$ for private federated learning of gradient boosting decision trees (GBDT). It supports running GBDT over both vertically and horizontally partitioned data. Vertical $\textsf {FederBoost}$ does not require any cryptographic operation and horizontal $\textsf {FederBoost}$ only requires lightweight secure aggregation. The key observation is that the whole training process of GBDT relies on the ordering of the data instead of the values. We fully implement $\textsf {FederBoost}$ and evaluate its utility and efficiency through extensive experiments performed on three public datasets. Our experimental results show that both vertical and horizontal $\textsf {FederBoost}$ achieve the same level of accuracy with centralized training where all data are collected in a central server; and they are 4-5 orders of magnitude faster than the state-of-the-art solutions for federated decision tree training; hence offering practical solutions for industrial applications.
- 16 May 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 21 (3) , 1274-1285
${\sf FederBoost}$: Private Federated Learning for GBDT
Federated Learning (FL) has been an emerging trend in machine learning and artificial intelligence. It allows multiple participants to collaboratively train a better global model and offers a privacy-aware paradigm for model training since it does not require participants to release their original training data. However, existing FL solutions for vertically partitioned data or decision trees require heavy cryptographic operations. In this paper, we propose a framework named $\textsf {FederBoost}$ for private federated learning of gradient boosting decision trees (GBDT). It supports running GBDT over both vertically and horizontally partitioned data. Vertical $\textsf {FederBoost}$ does not require any cryptographic operation and horizontal $\textsf {FederBoost}$ only requires lightweight secure aggregation. The key observation is that the whole training process of GBDT relies on the ordering of the data instead of the values. We fully implement $\textsf {FederBoost}$ and evaluate its utility and efficiency through extensive experiments performed on three public datasets. Our experimental results show that both vertical and horizontal $\textsf {FederBoost}$ achieve the same level of accuracy with centralized training where all data are collected in a central server; and they are 4-5 orders of magnitude faster than the state-of-the-art solutions for federated decision tree training; hence offering practical solutions for industrial applications.
- Open Access
"Adversarial Examples" for Proof-of-Learning
In S&P '21, Jia et al. proposed a new concept/mechanism named proof-of-learning (PoL), which allows a prover to demonstrate ownership of a machine learning model by proving integrity of the training procedure. It guarantees that an adversary cannot construct a valid proof with less cost (in both computation and storage) than that made by the prover in generating the proof. A PoL proof includes a set of intermediate models recorded during training, together with the corresponding data points used to obtain each recorded model. Jia et al. claimed that an adversary merely knowing the final model and training dataset cannot efficiently find a set of intermediate models with correct data points. In this paper, however, we show that PoL is vulnerable to ``adversarial examples''! Specifically, in a similar way as optimizing an adversarial example, we could make an arbitrarily-chosen data point ``generate'' a given model, hence efficiently generating intermediate models with correct data points. We demonstrate, both theoretically and empirically, that we are able to generate a valid proof with significantly less cost than generating a proof by the prover.
Open Access"Adversarial Examples" for Proof-of-Learning
In S&P '21, Jia et al. proposed a new concept/mechanism named proof-of-learning (PoL), which allows a prover to demonstrate ownership of a machine learning model by proving integrity of the training procedure. It guarantees that an adversary cannot construct a valid proof with less cost (in both computation and storage) than that made by the prover in generating the proof. A PoL proof includes a set of intermediate models recorded during training, together with the corresponding data points used to obtain each recorded model. Jia et al. claimed that an adversary merely knowing the final model and training dataset cannot efficiently find a set of intermediate models with correct data points. In this paper, however, we show that PoL is vulnerable to ``adversarial examples''! Specifically, in a similar way as optimizing an adversarial example, we could make an arbitrarily-chosen data point ``generate'' a given model, hence efficiently generating intermediate models with correct data points. We demonstrate, both theoretically and empirically, that we are able to generate a valid proof with significantly less cost than generating a proof by the prover.