Kun He
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
Research Interests
Kun He has a total of 151 co-authors. Here are the top co-authors by number of shared publications:
- Jing Chen
Wuhan University
70 shared publications - Ruiying Du
Wuhan University
60 shared publications - Cong Wu
School of Cyber Science and Engineering, Wuhan University, Wuhan, China
25 shared publications - Meng Jia
Key Laboratory of Aerospace Information Security and Trusted Computing, Ministry of Education, School of Cyber Science and Engineering, Wuhan University, Wuhan, China
15 shared publications Ruiying Du
Key Laboratory of Aerospace Information Security and Trusted Computing, Ministry of Education, School of Cyber Science and Engineering, Wuhan University, Wuhan, China
9 shared publications- Min Shi
Key Laboratory of Aerospace Information Security and Trusted Computing, Ministry of Education, School of Cyber Science and Engineering, Wuhan University, Wuhan, China
8 shared publications
Latest Publications
- Open Access
- 7 October 2026
- Published by Association for Computing Machinery (ACM)
- p. 2152-2164
Breaking the Isolation: Coordinated Multi-agent Fuzzing for Smart Contracts with Multi-dimensional Objective Learning
- 7 October 2026
- Published by Association for Computing Machinery (ACM)
- p. 2152-2164
Open AccessBreaking the Isolation: Coordinated Multi-agent Fuzzing for Smart Contracts with Multi-dimensional Objective Learning
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 23 (5) , 11716-11733
Fortress: Multi-Level Secure and Efficient Distributed Learning
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 23 (5) , 11716-11733
Fortress: Multi-Level Secure and Efficient Distributed Learning
- Open Access
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 5939-5950
Efficient and Differentially Private Federated LLM Fine-Tuning on Heterogeneous Clients
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 5939-5950
Open AccessEfficient and Differentially Private Federated LLM Fine-Tuning on Heterogeneous Clients
- 18 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 21 (15566013) , 3973-3988
N Truths and a Lie: Consistency-Based Backdoor Defense for Vertical Federated Learning
Recently, a growing number of backdoor attacks have been implemented against vertical federated learning (VFL). To carry out such attacks, it is possible for an attacker to manipulate its local data and model, which are not accessible to the defender. This asymmetry complicates the design of defense mechanisms, resulting in a lack of effective defenses against backdoor attacks in VFL. In this paper, we propose a novel backdoor defense mechanism tailored for VFL. Specifically, we observe that the malicious embeddings provided by an attacker are often inconsistent with the embeddings provided by honest VFL participants and thus are less predictable from the latter. On the basis of this observation, we introduce a latent masked autoencoder (LMAE) to assess the semantic consistency of the embeddings provided by different VFL participants. On the basis of the LMAE, we further develop an algorithm to identify attackers and enable backdoor-resistant predictions. To evaluate the proposed defense, we conduct experiments involving four baseline defenses, seven backdoor attacks, and five datasets (CIFAR-10, CINIC-10, Yahoo Answers, BHI, and BM) of different modalities. The results show that our defense is effective at identifying attackers and achieves high prediction accuracy. Furthermore, we demonstrate that the defense is robust against various attacks in a wide range of scenarios.
- 18 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 21 (15566013) , 3973-3988
N Truths and a Lie: Consistency-Based Backdoor Defense for Vertical Federated Learning
Recently, a growing number of backdoor attacks have been implemented against vertical federated learning (VFL). To carry out such attacks, it is possible for an attacker to manipulate its local data and model, which are not accessible to the defender. This asymmetry complicates the design of defense mechanisms, resulting in a lack of effective defenses against backdoor attacks in VFL. In this paper, we propose a novel backdoor defense mechanism tailored for VFL. Specifically, we observe that the malicious embeddings provided by an attacker are often inconsistent with the embeddings provided by honest VFL participants and thus are less predictable from the latter. On the basis of this observation, we introduce a latent masked autoencoder (LMAE) to assess the semantic consistency of the embeddings provided by different VFL participants. On the basis of the LMAE, we further develop an algorithm to identify attackers and enable backdoor-resistant predictions. To evaluate the proposed defense, we conduct experiments involving four baseline defenses, seven backdoor attacks, and five datasets (CIFAR-10, CINIC-10, Yahoo Answers, BHI, and BM) of different modalities. The results show that our defense is effective at identifying attackers and achieves high prediction accuracy. Furthermore, we demonstrate that the defense is robust against various attacks in a wide range of scenarios.
- 11 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 21 (15566013) , 2978-2991
MagLive: Robust Voice Liveness Detection on Smartphones Using Magnetic Pattern Changes
Voice authentication has been widely used on smartphones. However, it remains vulnerable to spoofing attacks, where the attacker replays recorded voice samples from authentic humans using loudspeakers to bypass the voice authentication system. In this paper, we present MagLive, a robust voice liveness detection scheme designed for smartphones to mitigate such spoofing attacks. MagLive leverages the differences in magnetic pattern changes generated by different speakers (i.e., humans or loudspeakers) when speaking for liveness detection, which are captured by the built-in magnetometer on smartphones. To extract effective and robust magnetic features, MagLive utilizes a TF-CNN-SAF model as the feature extractor, which includes a time-frequency convolutional neural network (TF-CNN) combined with a self-attention-based fusion (SAF) model. Supervised contrastive learning is then employed to achieve user-irrelevance, device-irrelevance, and content-irrelevance. MagLive imposes no additional burden on users and does not rely on active sensing or specialized hardware. We conducted comprehensive experiments with various settings to evaluate the security and robustness of MagLive. Our results demonstrate that MagLive effectively distinguishes between humans and attackers (i.e., loudspeakers), achieving an average balanced accuracy (BAC) of 99.01% and an equal error rate (EER) of 0.77%.
- 11 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 21 (15566013) , 2978-2991
MagLive: Robust Voice Liveness Detection on Smartphones Using Magnetic Pattern Changes
Voice authentication has been widely used on smartphones. However, it remains vulnerable to spoofing attacks, where the attacker replays recorded voice samples from authentic humans using loudspeakers to bypass the voice authentication system. In this paper, we present MagLive, a robust voice liveness detection scheme designed for smartphones to mitigate such spoofing attacks. MagLive leverages the differences in magnetic pattern changes generated by different speakers (i.e., humans or loudspeakers) when speaking for liveness detection, which are captured by the built-in magnetometer on smartphones. To extract effective and robust magnetic features, MagLive utilizes a TF-CNN-SAF model as the feature extractor, which includes a time-frequency convolutional neural network (TF-CNN) combined with a self-attention-based fusion (SAF) model. Supervised contrastive learning is then employed to achieve user-irrelevance, device-irrelevance, and content-irrelevance. MagLive imposes no additional burden on users and does not rely on active sensing or specialized hardware. We conducted comprehensive experiments with various settings to evaluate the security and robustness of MagLive. Our results demonstrate that MagLive effectively distinguishes between humans and attackers (i.e., loudspeakers), achieving an average balanced accuracy (BAC) of 99.01% and an equal error rate (EER) of 0.77%.
- 9 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 3886-3901
POWER: High-Throughput Blockchain Based on Computing Power Utilization
Proof-of-Work (PoW) based blockchain reaches consensus by solving computational puzzles, with only the winners generating blocks. In other words, most nodes waste their computing power, resulting in a mismatch between transaction throughput and system scale (i.e., the number of nodes). In this paper, we propose POWER, a high transaction throughput blockchain architecture that utilizes node computing power. Specifically, by introducing hook block and transaction block, we design a parallel structure to increase the utilization of the computing power of nodes that package and add transactions to the blockchain. In the parallel structure, the hook block hangs the transaction block and solves the problem of transaction redundancy. We also present a round interval confirmation mechanism to increase the utilization of the computing power of nodes that confirm transactions. We conduct experiments on POWER and the experiment results show that POWER has a 63% higher transaction throughput and a 51% lower transaction confirmation latency compared to other schemes. In particular, the transaction confirmation accuracy of POWER is 5.9 times better than that of OHIE.
- 9 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 3886-3901
POWER: High-Throughput Blockchain Based on Computing Power Utilization
Proof-of-Work (PoW) based blockchain reaches consensus by solving computational puzzles, with only the winners generating blocks. In other words, most nodes waste their computing power, resulting in a mismatch between transaction throughput and system scale (i.e., the number of nodes). In this paper, we propose POWER, a high transaction throughput blockchain architecture that utilizes node computing power. Specifically, by introducing hook block and transaction block, we design a parallel structure to increase the utilization of the computing power of nodes that package and add transactions to the blockchain. In the parallel structure, the hook block hangs the transaction block and solves the problem of transaction redundancy. We also present a round interval confirmation mechanism to increase the utilization of the computing power of nodes that confirm transactions. We conduct experiments on POWER and the experiment results show that POWER has a 63% higher transaction throughput and a 51% lower transaction confirmation latency compared to other schemes. In particular, the transaction confirmation accuracy of POWER is 5.9 times better than that of OHIE.
- 28 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 21 (15566013) , 2549-2563
Realhybrid: A Hybrid Blockchain Consensus With Node-Level Switching
Blockchain consensus can be divided into synchronous consensus and asynchronous consensus according to the network status. In a real network environment, the network status of each node is constantly fluctuating. Hybrid consensus schemes adapt to network fluctuations through switching consensus protocol between asynchronous and synchronous. However, existing schemes are system-level switching, resulting in low efficiency. In this paper, we present Realhybrid, a hybrid consensus scheme with node-level switching, which enables every node to select appropriate consensus protocols based on their network status. We design corresponding protocols for each node to achieve efficient consensus under network fluctuations. Moreover, we establish a Realhybrid network model and quantify the relationship between its performance and system parameters. We conduct experiments on Realhybrid and the results show that it has 29% lower transaction waiting volume and 17% lower transaction confirmation latency compared to other hybrid consensus schemes.
- 28 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 21 (15566013) , 2549-2563
Realhybrid: A Hybrid Blockchain Consensus With Node-Level Switching
Blockchain consensus can be divided into synchronous consensus and asynchronous consensus according to the network status. In a real network environment, the network status of each node is constantly fluctuating. Hybrid consensus schemes adapt to network fluctuations through switching consensus protocol between asynchronous and synchronous. However, existing schemes are system-level switching, resulting in low efficiency. In this paper, we present Realhybrid, a hybrid consensus scheme with node-level switching, which enables every node to select appropriate consensus protocols based on their network status. We design corresponding protocols for each node to achieve efficient consensus under network fluctuations. Moreover, we establish a Realhybrid network model and quantify the relationship between its performance and system parameters. We conduct experiments on Realhybrid and the results show that it has 29% lower transaction waiting volume and 17% lower transaction confirmation latency compared to other hybrid consensus schemes.
- 2 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 21 (15566013) , 741-756
MsgFilter: Proactive Anti-Harassment Sender-Anonymous Messaging System
Anonymous submissions inspire people to speak up since hiding their identities can protect them from negative influence by their own words. However, the abuse of anonymity may bring harassment to those public submission callers. Existing works only handle DoS attacks or block harassment senders in an active manner, which behave poorly in the early prevention of uncharacterized harassment. In this paper, we propose MsgFliter, a sender-anonymous messaging system with proactive anti-harassment mechanism. Our core idea is to prevent unanswered senders from sending messages continually while keeping their identities, messages, and sender types secret. To meet the functionality and security requirements of MsgFliter, we propose the Anti-Harassment Anonymous Authentication (AHAA) protocol. We associate messages from the same sender through linkable tags and invalidate linkability when a message is replied to. To achieve session indistinguishability, we further combine the proposed anonymous authentication with zero-knowledge proofs of disjunctive relations. We implement MsgFliter and compare its performance with related solutions. Experimental results show that our solution is practicable.
- 2 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 21 (15566013) , 741-756
MsgFilter: Proactive Anti-Harassment Sender-Anonymous Messaging System
Anonymous submissions inspire people to speak up since hiding their identities can protect them from negative influence by their own words. However, the abuse of anonymity may bring harassment to those public submission callers. Existing works only handle DoS attacks or block harassment senders in an active manner, which behave poorly in the early prevention of uncharacterized harassment. In this paper, we propose MsgFliter, a sender-anonymous messaging system with proactive anti-harassment mechanism. Our core idea is to prevent unanswered senders from sending messages continually while keeping their identities, messages, and sender types secret. To meet the functionality and security requirements of MsgFliter, we propose the Anti-Harassment Anonymous Authentication (AHAA) protocol. We associate messages from the same sender through linkable tags and invalidate linkability when a message is replied to. To achieve session indistinguishability, we further combine the proposed anonymous authentication with zero-knowledge proofs of disjunctive relations. We implement MsgFliter and compare its performance with related solutions. Experimental results show that our solution is practicable.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 23 (5) , 1-13
Silent Poisoning: A Stealthy and Unified Dual-Space Framework for VFL Backdoor Attacks
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 23 (5) , 1-13
Silent Poisoning: A Stealthy and Unified Dual-Space Framework for VFL Backdoor Attacks