Zhenwan Peng

Zhenwan Peng

Auto-generated profile, based on publicly available information

Zhenwan Peng

Anhui Medical University

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
16 March 2023 in Mathematics
Verified
Journal Article
Open access
PDF file
YWYu Wang
Yu Wang
YWYuan Wang
Yuan Wang
ZPZhenwan Peng
Zhenwan Peng
FZFeifan Zhang
Feifan Zhang
FYFei YANG
Fei YANG
Relation extraction, a fundamental task in natural language processing, aims to extract entity triples from unstructured data. These triples can then be used to build a knowledge graph. Recently, pre-training models that have learned prior semantic and syntactic knowledge, such as BERT and ERNIE, have enhanced the performance of relation extraction tasks. However, previous research has mainly focused on sequential or structural data alone, such as the shortest dependency path, ignoring the fact that fusing sequential and structural features may improve the classification performance. This study proposes a concise approach using the fused features for the relation extraction task. Firstly, for the sequential data, we verify in detail which of the generated representations can effectively improve the performance. Secondly, inspired by the pre-training task of next-sentence prediction, we propose a concise relation extraction approach based on the fusion of sequential and structural features using the pre-training model ERNIE. The experiments were conducted on the SemEval 2010 Task 8 dataset and the results show that the proposed method can improve the F1 value to 0.902.

15 April 2024 in International Journal of Intelligent Systems
Verified
Journal Article
Open access
PDF file
YWYu Wang
Yu Wang
LZLuyao Zhou
Luyao Zhou
YWYuan Wang
Yuan Wang
ZPZhenwan Peng
Zhenwan Peng
Pretrained Language Models (PLMs) acquire rich prior semantic knowledge during the pretraining phase and utilize it to enhance downstream Natural Language Processing (NLP) tasks. Entity Matching (EM), a fundamental NLP task, aims to determine whether two entity records from different knowledge bases refer to the same real-world entity. This study, for the first time, explores the potential of using a PLM to boost the EM task through two transfer learning techniques, namely, fine-tuning and prompt learning. Our work also represents the first application of the soft prompt in an EM task. Experimental results across eleven EM datasets show that the soft prompt consistently outperforms other methods in terms of F1 scores across all datasets. Additionally, this study also investigates the capability of prompt learning in few-shot learning and observes that the hard prompt achieves the highest F1 scores in both zero-shot and one-shot context. These findings underscore the effectiveness of prompt learning paradigms in tackling challenging EM tasks.

22 May 2024 in The Astronomical Journal
Verified
Journal Article
Open access
FZFeifan Zhang
Feifan Zhang
BCBaolong Chen
Baolong Chen
YWYu Wang
Yu Wang
ZPZhenwan Peng
Zhenwan Peng
ZZZengxiang Zhou
Zengxiang Zhou
JWJianping Wang
Jianping Wang
The Large Sky Area Multi-Object Fiber Spectroscopy Telescope (LAMOST) is one of the most effective multiobject spectroscopic instruments. Its survey efficiency is guaranteed by simultaneously positioning multiple fibers via 4000 robotic fiber positioners (RFPs). With the further updates to LAMOST, the new-generation RFPs will be smaller, and the number of RFPs will increase to 5000. The RFPs are densely packed with shared working space. Thus, they may collide with each other, leading to them damaged and reducing the survey speed. In this study, we propose a new motion planning algorithm that prevents the collision of RFPs. To simplify the collision avoidance problem, we transform the motion planning process from a dynamic one into a static one by selecting one of the RFPs in each collision pair as the waiting robot. Accordingly, we design a method for choosing the waiting robot, and use a rapidly exploring random tree to plan a collision-free path for the waiting robot. However, there may be blocks between the waiting robots and their neighbors. Therefore, we also design methods to resolve these blocks. Simulations suggest that the proposed algorithm can prevent 98.4% of the collisions. About 99.9% of the positioners can reach their targets without collisions. Although developed for LAMOST, we believe that our algorithm can also be used for other instruments with equal-arm theta-phi positioners, such as the Dark Energy Spectroscopic Instrument.

14 March 2024 in Physica Scripta
Verified
Journal Article
Open access
PDF file
ZPZhen-Wan Peng
Zhen-Wan Peng
RSRunhua Shi
Runhua Shi
RDRan Ding
Ran Ding
FZFeifan Zhang
Feifan Zhang
Secure multiparty computational geometry is an essential field of secure multiparty computation. It involves solving geometric problems without disclosing any private geometric information from each participating party. Privacy-preserving point-line relation determination, a pivotal branch of secure multiparty computational geometry, holds significant potential for applications in various domains, including business, engineering, and military. In this paper, leveraging the assistance of a third party, we introduce a quantum solution to privacy-preserving point-line relation determination. This approach incorporates quantum key distribution and one-time pad encryption. Under the condition that the third party does not conspire with any participant, our protocol ensures heightened security and enhanced privacy protection compared to classical methods. This superiority is attributed to the underlying principles of quantum mechanics and the utilization of one-time pad encryption. Furthermore, we extend our protocol to involve multiple third parties, employing the technology of quantum secret sharing. In scenarios where these third parties refrain from colluding with any single participant, no information is divulged to the other participant. Consequently, the protocol becomes more rational and secure when multiple third parties are involved.

21 October 2024 in IEEE Transactions on Emerging Topics in Computing
Verified
Journal Article
Closed access
YHYunfei He
Yunfei He
YWYang Wu
Yang Wu
LHLishan Huang
Lishan Huang
ZPZhenwan Peng
Zhenwan Peng
FYFei YANG
Fei YANG
VSVictor Sheng
Victor Sheng
Empowered by their remarkable advantages, graph neural networks (GNN) serve as potent tools for embedding graph-structured data and finding applications across various domains. Particularly, a prevalent assumption in most GNNs is the reliability of the underlying graph structure. This assumption, often implicit, can inadvertently lead to the propagation of misleading information through structures like false links. In response to this challenge, numerous methods for graph structure learning (GSL) have been developed. Among these methods, one popular approach is to construct a simple and intuitive K-nearest neighbor (KNN) graph as a sample to infer true graph structure. However, KNN graphs that follow the single-point distribution can easily mislead the true graph structure estimation. The primary reason is that, from a statistical perspective, the KNN graph, as a sample, follows a single-point distribution, whereas the true graph structure, as the population, as a whole mostly follows a long-tail distribution. In theory, the sample and the population should share the same distribution; otherwise, accurately inferring the true graph structure becomes challenging. To address this problem, this paper proposes an Adaptive Graph Structure Estimation with Long-Tail Distributed Implicit Graph, referred to as AGSEI. AGSEI comprises three main components: long-tail implicit graph construction, explicit graph structure estimation, and joint optimization. The first component relies on a multi-layer graph convolutional network to learn low-order to high-order node representations, compute node similarity, and construct several corresponding long-tail implicit graphs. Since the original imperfect graph structure can mislead GNNs into propagating false information, it reduces the reliability of the long-tail implicit graphs. AGSEI attempts to limit the aggregation of irrelevant information by introducing the Hilbert-Schmidt independence criterion. That is, maximizing the dependence between the predicted label and ground truth. With this strategy, AGSEI can learn node features dependent on labels to facilitate the construction of reliable long-tail implicit graphs, and then provide adaptive multi-view graph structure information to support subsequent GSL. In the second component, the graph structure is estimated using the stochastic block model (SBM) with the Expectation-Maximization algorithm. Considering that it is difficult for a single GSL to approach the true graph structure, the third part considers the joint optimization of the long-tail implicit graph construction and the explicit graph structure estimation. This involves optimizing the two parts alternately until the model converges. We conducted multiple experiments on five public datasets, including tasks such as classification and clustering. These experiments not only demonstrated the performance of AGSEI but also confirmed that the graph structures it estimates align with the long-tail distribution.

3 February 2026 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
LYLongfeng Yu
Longfeng Yu
ZDZhaoru Dong
Zhaoru Dong
YWYu Wang
Yu Wang
ZPZhenwan Peng
Zhenwan Peng
FZFeifan Zhang
Feifan Zhang
Needle puncture serves as the foundation for a broad range of minimally invasive surgical diagnostic and therapeutic procedures. Flexible needles have the potential to overcome the limitations of traditional rigid needles, particularly their inability to bypass obstacles and correct needle tip deviations, making them a promising alternative. However, when the needle hub rotates during insertion, torsional interactions between the needle shaft and surrounding tissue result in a discrepancy between the rotation angles of the needle hub and the needle tip. This discrepancy can lead to needle tip misalignment, thereby affecting patient prognosis and compromising diagnostic reliability. Furthermore, accurately measuring the orientation of the needle tip remains a challenge. To address these issues, we propose a method for estimating and compensating for orientation discrepancies, which improves puncture accuracy, enhances patient prognosis, and boosts diagnostic reliability. Specifically, based on puncture images, we introduce a projection-based approach to estimate the actual rotation angle of the needle tip, and construct a dataset containing the corresponding rotation angles of the needle hub and tip. We then utilize a neural network to estimate the true orientation of the needle tip from a specific needle hub rotation angle. Experimental results demonstrate that the proposed method can effectively estimate and compensate for the discrepancy between the orientations of the needle tip and hub, reducing the average rotation error from 9.6° to 0.6°. Additionally, puncture experiments conducted using paths planned via the particle swarm optimization algorithm verify that the method improves puncture precision, decreasing the error from 2.42±0.65 mm to 0.96±0.65 mm. These results indicate that the method is feasible and effective for enhancing patient prognosis and diagnostic reliability.