Rongyuan Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Rongyuan Chen has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
- Lizhi Shen
Hunan University of Technology and Business, China
3 shared publications - Shaonian Huang
School of Information Science and Engineering, Central South University, Changsha 410083, China
3 shared publications - Zhixiong He
Hunan University of Technology and Business, Hunan Changsha 410205, China
2 shared publications - Jiebin Wen1 shared publication
- Lang Chen
Hunan University of Technology and Business
1 shared publication - Xiancheng Zhou
School of Intelligent Engineering and Intelligent Manufacturing, Hunan University of Technology and Business, Changsha 410205, China
1 shared publication
Latest Publications
- Open Access
- 7 March 2026
- Published by Elsevier in SSRN Electronic Journal
ST-AHM: A Soft-Threshold Sparse-Attention Hypergraph Model with Contrastive Learning for Session-based Recommendation
Session-based recommendation (SBR) provides personalized recommendations by leveraging anonymous user click data. The key challenge in SBR lies in effectively utilizing information within sessions while minimizing the impact of irrelevant data on prediction performance. To address this, we propose a novel Soft-Threshold-Attention Hypergraph Model (ST-AHM). First, ST-AHM employs two parallel channels to model local and global relationships separately. For local relationships, the model constructs a hypergraph structure to represent user click sequences. It extracts local features via a gated hypergraph neural network and refines them using a soft-threshold sparse multi-head attention mechanism to reduce noise, thereby obtaining accurate session target representations. For global relationships, relational group graph convolution is employed to aggregate global information, and a global sparse attention mechanism is applied to further refine the representation of global features. Finally, ST-AHM comprehensively utilizes contrastive learning to integrate effective information from both channels, thereby enhancing the model's ability to understand user intent. Experiments demonstrate that on three benchmark datasets—Tmall, Nowplaying, and 30musics—ST-AHM outperforms mainstream models such as MSGAT, MLEUP, and TSREC, achieving average improvements of 19.23%, 12.71%, 23.91%, and 24.86% in the P@10, P@20, MRR@10, and MRR@20 performance metrics, respectively. The results highlight the effectiveness and value of simultaneously modeling intra-session and inter-session information, fully leveraging both local and global information while mitigating the impact of noise.
- 7 March 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessST-AHM: A Soft-Threshold Sparse-Attention Hypergraph Model with Contrastive Learning for Session-based Recommendation
Session-based recommendation (SBR) provides personalized recommendations by leveraging anonymous user click data. The key challenge in SBR lies in effectively utilizing information within sessions while minimizing the impact of irrelevant data on prediction performance. To address this, we propose a novel Soft-Threshold-Attention Hypergraph Model (ST-AHM). First, ST-AHM employs two parallel channels to model local and global relationships separately. For local relationships, the model constructs a hypergraph structure to represent user click sequences. It extracts local features via a gated hypergraph neural network and refines them using a soft-threshold sparse multi-head attention mechanism to reduce noise, thereby obtaining accurate session target representations. For global relationships, relational group graph convolution is employed to aggregate global information, and a global sparse attention mechanism is applied to further refine the representation of global features. Finally, ST-AHM comprehensively utilizes contrastive learning to integrate effective information from both channels, thereby enhancing the model's ability to understand user intent. Experiments demonstrate that on three benchmark datasets—Tmall, Nowplaying, and 30musics—ST-AHM outperforms mainstream models such as MSGAT, MLEUP, and TSREC, achieving average improvements of 19.23%, 12.71%, 23.91%, and 24.86% in the P@10, P@20, MRR@10, and MRR@20 performance metrics, respectively. The results highlight the effectiveness and value of simultaneously modeling intra-session and inter-session information, fully leveraging both local and global information while mitigating the impact of noise.
- Open Access
- 19 January 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-9
Research On Palmprint Recognition Based On Mechanism And Data
Among the existing palmprint recognition technologies, traditional methods cannot adaptively extract multi-level features of palmprint images to better represent image information, and methods based on deep learning fail to make full use of prior knowledge existing in palmprint images. To solve this problem, A palmprint recognition method was proposed based on fusion mechanism and data, and constructs a palmprint recognition data model based on deep learning. The prior knowledge in palmprint images is extracted and added to the data model as mechanism knowledge to improve the model's ability to discriminate palmprint features. The multi-direction Gabor filter is used to optimize the prediction results and improve the recognition performance. The experimental results show that the recognition accuracy of the proposed method on CASIA palmprint database is 99.41%, which is better than the existing mainstream methods.
- 19 January 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-9
Open AccessResearch On Palmprint Recognition Based On Mechanism And Data
Among the existing palmprint recognition technologies, traditional methods cannot adaptively extract multi-level features of palmprint images to better represent image information, and methods based on deep learning fail to make full use of prior knowledge existing in palmprint images. To solve this problem, A palmprint recognition method was proposed based on fusion mechanism and data, and constructs a palmprint recognition data model based on deep learning. The prior knowledge in palmprint images is extracted and added to the data model as mechanism knowledge to improve the model's ability to discriminate palmprint features. The multi-direction Gabor filter is used to optimize the prediction results and improve the recognition performance. The experimental results show that the recognition accuracy of the proposed method on CASIA palmprint database is 99.41%, which is better than the existing mainstream methods.
- Open Access
- 30 August 2022
- Published by Wiley in Computational and Mathematical Methods in Medicine
- Vol. 2022, 1-12
Optimal Modeling of Anti-Breast Cancer Candidate Drugs Based on Graph Model Feature Selection
Breast cancer is one of the most widespread and fatal cancers in women. At present, anticancer drug-inhibiting estrogen receptor subtype (ER) can greatly improve the cure rate for breast cancer patients, so the research and development of this kind of drugs are very urgent. In this paper, the problem of how to screen excellent anticancer drugs is abstracted as an optimization problem. Firstly, the graph model is used to extract low-dimensional features with strong distinguishing and describing ability according to various attributes of candidate compounds, and then, kernel functions are used to map these features to high-dimensional space. Then, the quantitative analysis model of ER biological activity and the classification model based on ADMET properties of the support vector machine are constructed. Finally, sequential least square programming (SLSQP) is utilized to solve the ER biological activity model. The experimental results show that for anticancer data sets, compared with principal component analysis (PCA), the error rate of the graph model constructed in this paper is reduced by 6.4%, 15%, and 7.8% on mean absolute error (MAE), mean squared error (MSE), and root mean square error (RMSE), respectively. In terms of classification prediction, compared with principal component analysis (PCA), the recall and precision rates of this method are enhanced by 19.5% and 12.41%, respectively. Finally, the optimal biological activity value (IC50_nM) 34.6 and inhibitory biological activity value (pIC50) 7.46 were obtained.
- 30 August 2022
- Published by Wiley in Computational and Mathematical Methods in Medicine
- Vol. 2022, 1-12
Open AccessOptimal Modeling of Anti-Breast Cancer Candidate Drugs Based on Graph Model Feature Selection
Breast cancer is one of the most widespread and fatal cancers in women. At present, anticancer drug-inhibiting estrogen receptor subtype (ER) can greatly improve the cure rate for breast cancer patients, so the research and development of this kind of drugs are very urgent. In this paper, the problem of how to screen excellent anticancer drugs is abstracted as an optimization problem. Firstly, the graph model is used to extract low-dimensional features with strong distinguishing and describing ability according to various attributes of candidate compounds, and then, kernel functions are used to map these features to high-dimensional space. Then, the quantitative analysis model of ER biological activity and the classification model based on ADMET properties of the support vector machine are constructed. Finally, sequential least square programming (SLSQP) is utilized to solve the ER biological activity model. The experimental results show that for anticancer data sets, compared with principal component analysis (PCA), the error rate of the graph model constructed in this paper is reduced by 6.4%, 15%, and 7.8% on mean absolute error (MAE), mean squared error (MSE), and root mean square error (RMSE), respectively. In terms of classification prediction, compared with principal component analysis (PCA), the recall and precision rates of this method are enhanced by 19.5% and 12.41%, respectively. Finally, the optimal biological activity value (IC50_nM) 34.6 and inhibitory biological activity value (pIC50) 7.46 were obtained.
- 1 January 2022
- Published by Springer Nature
- p. 1095-1106
Research on Splicing of Horizontal and Longitudinal Shredded Paper Based on Hungarian Algorithm
- 1 January 2022
- Published by Springer Nature
- p. 1095-1106
Research on Splicing of Horizontal and Longitudinal Shredded Paper Based on Hungarian Algorithm