HZHongwei Zhou

Hongwei Zhou

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Hongwei Zhou has a total of 17 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    MIRAGE: manipulation-immune RAG with adaptive guard against textual evidence attacks

    by Qiao Yuan, Feiyang Li, Hao Hu, Yanming Mu, Hongwei Zhou, Yuling Liu, Yuchen Zhang

    Retrieval-augmented generation (RAG) substantially improves large language models (LLMs) on knowledge-intensive tasks by incorporating external knowledge, yet open knowledge sources introduce prominent security risks. The SafeRAG benchmark identifies four novel textual evidence manipulation attacks, namely silver noise, inter-context conflict, soft advertisement, and White Denial-of-Service (White DoS), which bypass conventional defense components and threaten both the security and availability of RAG systems. Existing single-stage defenses fail to capture the implicit malicious characteristics of these attacks, while multi-stage frameworks are limited by fixed parameters, weak conflict resolution, and insufficient detection of implicit toxicity and availability threats. To address these limitations, we propose MIRAGE, a two-stage adaptive defense framework combining retrieval-side purification and generation-side resolution. The first stage applies attack-characteristic-aware dynamic K-means clustering and dynamic ROUGE-L threshold filtering to eliminate malicious documents. The second stage employs attack-adaptive knowledge integration rules alongside an enhanced self-evaluation mechanism that jointly verifies factual accuracy, implicit toxicity, and availability. Under the static, non-adaptive attack setting defined by SafeRAG, experiments on DeepSeek-v3.2 show that MIRAGE achieves average improvements of 20.9% in Attack Failure Rate (AFR), 11.5% in Retrieval Accuracy (RA), and 14.1% in F1(avg) over baselines across all four attack scenarios, providing effective technical support for the stable deployment of RAG systems in complex adversarial environments.

  • Open Access

    FlanBC: A Semantic-Structural Sequence Labeling Framework for Log Parsing

    by Jinhui Yuan, Bin Guan, Kun Wen, Jiawei Fang, Hongwei Zhou

    Log parsing converts raw system logs into structured templates and is a key preprocessing step for Artificial Intelligence for IT Operations (AIOps). Existing parsers face a practical trade-off: rule-based methods offer high throughput but limited adaptability across heterogeneous log sources, whereas Large Language Model (LLM)-based parsers achieve broader semantic coverage at the cost of inference latency, privacy exposure, and cloud dependency. This paper presents FlanBC, a log parsing framework that formulates template extraction as a BIO (Beginning, Inside, Outside) sequence-labeling task and integrates a Flan-T5 semantic encoder, Bidirectional Long Short-Term Memory (BiLSTM) layers for local sequential modeling, and a Conditional Random Field (CRF) decoder for structured label prediction. Log-specific preprocessing and a subword-to-token alignment mechanism adapt the general-purpose encoder to semi-structured log data. A layer-freezing strategy reduces the number of parameters updated during training. The framework supports local inference without external API dependency. Experiments on three benchmark datasets from LogHub (HDFS, BGL, OpenStack) under a supervised random-split setup evaluate parsing accuracy, training efficiency, statistical stability across random seeds, and component contributions. FlanBC achieves a Group Accuracy of 99.32% on HDFS and 98.47% on BGL, with an inference throughput of 700+ logs/s on a consumer-grade GPU. On OpenStack, performance is lower (GA = 92.54%), reflecting the challenge that diverse natural-language-like logs pose for compact encoder-based models. Under a stricter template-disjoint split that prevents template overlap between training and test sets, FlanBC achieves an average Group Accuracy of 91.14%, indicating that the model generalizes to unseen templates beyond in-distribution recognition. Ablation results indicate that the semantic encoder, BiLSTM module, and CRF decoder each contribute to prediction accuracy. These findings suggest that domain-adapted semantic encoders combined with structured decoding offer a practical accuracy–efficiency balance for log parsing in settings where local, cloud-free inference is preferred.

  • Open Access

    DLogParser: An Efficient Dynamic Log Parser with Multiple Grouping Criteria

    by Jinhui Yuan, Chao Wang, Hongwei Zhou, Yucheng Zhang, Yongwei Wang

    Most existing log parsers are static. When parsing logs with a static parser, the accuracy tends to fluctuate significantly. To overcome this issue, this paper proposes a dynamic log parser named DLogParser. The core idea of DLogParser is to select different parsing policies based on log features. DLogParser first parses a small batch of sample log messages, then analyzes log characteristics from the parsing results, and determines an appropriate parsing policy for the current logs. Then it parses all remaining logs according to the determined policy. To support dynamic parsing policies, DLogParser incorporates 5 grouping criteria for log features, including length, punctuation, first token, last token, and key token, and establishes 7 rules for parsing policy generation. We evaluated DLogParser on public datasets from LogHub. The experimental results demonstrate that compared to 11 existing log parsers, DLogParser achieves an accuracy of 90.3% with an acceptable performance loss.

  • Open Access

    Cluster Head Monitoring Nodes Secret Selection Protocol in Sensor Networks

    by Jinhui Yuan, Hongwei Zhou

    In sensor networks, as the data collection point of nodes in the cluster, cluster head is easy to be the key target of attack. Once the cluster head node is controlled by the attacker, it will have a fatal impact on the security of the network. In order to solve the above problem, this paper proposes a cluster head monitoring nodes secret selection protocol. It uses secret selection algorithm to select the cluster head monitoring node from the neighbor nodes of the cluster head node. Thus, the attacker can not confirm the monitoring node, it results that the cluster head node can not be attacked. The protocol uses the spatio-temporal correlation between cluster head and cluster head monitoring nodes. Once it finds that cluster head node is suspected anomaly, it reports immediately to sink node. The analysis of protocol security shows that the protocol can meet the energy efficiency of sensor network and ensure the security of the system.

  • Open Access

    Remote Attestation of Sensor Nodes on Node Personalization: One Case Study

    by Jinhui Yuan, Hongwei Zhou, Zhixin Wang

    For the verifier has to identify the attester which introduces more computation, homogenous node is obstacle for remote attestation of sensor nodes. In this paper, we firstly propose node personalization, which means that every node is different from others in the sensor network, to save the computation energy. Then, we discuss the principle of node personalization. Lastly, we present a practical case. More details, we develop node personalization with obfuscated code, and verify code integrity with cascading hashing. Our discussion shows the solution we present is feasible.

  • Protect sensitive data with lightweight memory encryption

    by Hongwei Zhou, Jinhui Yuan, Rui Xiao, Kai Zhang, Jingyao Sun

    Since current commercial processor is not able to deal with the data in the cipher text, the sensitive data have to be exposed in the memory. It leaves a window for the adversary. To protect the sensitive data, a direct idea is to encrypt the data when the processor does not access them. On the observation, we have developed a lightweight memory encryption, called LeMe, to protect the sensitive data in the application. LeMe marks the sensitive data in the memory with the page table entry, and encrypts the data in their free time. LeMe is built on the Linux with a 3.17.6 kernel, and provides four user interfaces as dynamic link library. Our evaluations show LeMe is effective to protect the sensitive data and incurs an acceptable performance overhead.