Yingqi Zhao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Interests
Yingqi Zhao has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Aliaksandr Hubarevich
Istituto Italiano di Tecnologia, Genova, Italy
1 shared publication - Changli Nie
College of CS, TJ Key Lab of NDST, Nankai University, Tianjin, China
1 shared publication - Francesco De Angelis
Division of General Surgery & Bariatric Center of Excellence IFSO-EC, Department of Medico-Surgical Sciences and Biotechnologies, University La Sapienza of Rome, Latina, Italy
1 shared publication - Haiwei Zhang
College of Computer Science, Nankai University, Tianjin, China
1 shared publication - Jian-An Huang
Research Unit of Health Sciences and Technology, Faculty of Medicine, University of Oulu, Aapistie 5 A, 90220 Oulu, Finland
1 shared publication - Jin Jin
School of Information and Engineering, Zhengzhou University, Zhengzhou, China
1 shared publication
Latest Publications
- Open Access
- 10 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 914-923
DHMAE: A Disentangled Hypergraph Masked Autoencoder for Group Recommendation
Group recommendation aims to suggest items to a group of users that are suitable for the group. Although some existing powerful deep learning models have achieved improved performance, various aspects remain unexplored: (1) Most existing models using contrastive learning tend to rely on high-quality data augmentation which requires precise contrastive view generation; (2) There is multifaceted natural noise in group recommendation, and additional noise is introduced during data augmentation; (3) Most existing hypergraph neural network-based models over-entangle the information of members and items, ignoring their unique characteristics. In light of this, we propose a highly effective Disentangled Hypergraph Masked Auto Encoder-enhanced method for group recommendation (DHMAE), combining a disentangled hypergraph neural network with a graph masked autoencoder. This approach creates self-supervised signals without data augmentation by masking the features of some nodes and hyperedges and then reconstructing them. For the noise problem, we design a masking strategy that relies on pre-computed degree-sensitive probabilities for the process of masking features. Furthermore, we propose a disentangled hypergraph neural network for group recommendation scenarios to extract common messages of members and items and disentangle them during the convolution process. Extensive experiments demonstrate that our method significantly outperforms state-of-the-art models and effectively addresses the noise issue.
- 10 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 914-923
Open AccessDHMAE: A Disentangled Hypergraph Masked Autoencoder for Group Recommendation
Group recommendation aims to suggest items to a group of users that are suitable for the group. Although some existing powerful deep learning models have achieved improved performance, various aspects remain unexplored: (1) Most existing models using contrastive learning tend to rely on high-quality data augmentation which requires precise contrastive view generation; (2) There is multifaceted natural noise in group recommendation, and additional noise is introduced during data augmentation; (3) Most existing hypergraph neural network-based models over-entangle the information of members and items, ignoring their unique characteristics. In light of this, we propose a highly effective Disentangled Hypergraph Masked Auto Encoder-enhanced method for group recommendation (DHMAE), combining a disentangled hypergraph neural network with a graph masked autoencoder. This approach creates self-supervised signals without data augmentation by masking the features of some nodes and hyperedges and then reconstructing them. For the noise problem, we design a masking strategy that relies on pre-computed degree-sensitive probabilities for the process of masking features. Furthermore, we propose a disentangled hypergraph neural network for group recommendation scenarios to extract common messages of members and items and disentangle them during the convolution process. Extensive experiments demonstrate that our method significantly outperforms state-of-the-art models and effectively addresses the noise issue.
- 21 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 53-59
D2D Communication Power Control Based on Deep Q Learning and Fractional Frequency Reuse
Underlaying Device-to-Device (D2D) communication in 5G cellular networks can improve spectrum utilization and reduce communication latency by reusing cellular networks spectrum resources. However, D2D communication can cause interference to the cellular networks. To address the problem of mutual interference between cellular user equipment (CUE) and D2D user equipment (DUE), a deep Q learning (DQL) combined with fractional frequency reuse (FFR) algorithm is proposed. Firstly, FFR is implemented to reasonably allocate frequency band resources for all users, and then the DQL algorithm is used for power control to achieve the total sum rate maximization. Simulation results show that the proposed DQL algorithm effectively improves the sum rate compared to several traditional power control algorithms, and exhibits better generalization ability and lower time overhead.
- 21 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 53-59
D2D Communication Power Control Based on Deep Q Learning and Fractional Frequency Reuse
Underlaying Device-to-Device (D2D) communication in 5G cellular networks can improve spectrum utilization and reduce communication latency by reusing cellular networks spectrum resources. However, D2D communication can cause interference to the cellular networks. To address the problem of mutual interference between cellular user equipment (CUE) and D2D user equipment (DUE), a deep Q learning (DQL) combined with fractional frequency reuse (FFR) algorithm is proposed. Firstly, FFR is implemented to reasonably allocate frequency band resources for all users, and then the DQL algorithm is used for power control to achieve the total sum rate maximization. Simulation results show that the proposed DQL algorithm effectively improves the sum rate compared to several traditional power control algorithms, and exhibits better generalization ability and lower time overhead.
- Open Access
Nanofluidic trapping and enhanced Raman detection of single biomolecules in plasmonic bowl-shaped nanopore
Solid-state nanopores are emerging platforms for single-molecule protein sequencing due to their tolerance to hash physiology environment and compatibility with different electrical and optical detection methods. However, they suffer from poor molecular manipulations that were twisted with and thus limited by the detection methods. Here, we report a bowl-shaped plasmonic gold nanopore on silicon nitride with hydrogel to demonstrate near-field nanofluidic manipulation of DNA translocation for plasmon-enhanced Raman spectroscopic detection. The hydrogel linearized the DNA, and the linear DNA was trapped in the nanopore for tens of seconds due assumably to bipolar effect of the nanopore that generate electroosmotic sheath flow and bipolar surface charge distribution. Their combination led to a near-field confinement of the DNA in the nanopore hot spot to allow stable Raman detection. We envision that a combination of Raman spectroscopy with the bowl-shaped nanopores can succeed in single-molecule protein sequencing in a label-free way
Open AccessNanofluidic trapping and enhanced Raman detection of single biomolecules in plasmonic bowl-shaped nanopore
Solid-state nanopores are emerging platforms for single-molecule protein sequencing due to their tolerance to hash physiology environment and compatibility with different electrical and optical detection methods. However, they suffer from poor molecular manipulations that were twisted with and thus limited by the detection methods. Here, we report a bowl-shaped plasmonic gold nanopore on silicon nitride with hydrogel to demonstrate near-field nanofluidic manipulation of DNA translocation for plasmon-enhanced Raman spectroscopic detection. The hydrogel linearized the DNA, and the linear DNA was trapped in the nanopore for tens of seconds due assumably to bipolar effect of the nanopore that generate electroosmotic sheath flow and bipolar surface charge distribution. Their combination led to a near-field confinement of the DNA in the nanopore hot spot to allow stable Raman detection. We envision that a combination of Raman spectroscopy with the bowl-shaped nanopores can succeed in single-molecule protein sequencing in a label-free way