Wei Xu
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
Wei Xu has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Lixiang Xu
College of Artificial Intelligence Engineering, Hefei Institute of Technology, Hefei 238076, China
3 shared publications - Qingzhe Cui
School of Computer Engineering & Science
3 shared publications - Yuan Yan Tang
Faculty of Science and Technology, University of Macau, Macau, China
2 shared publications - Biao Zhou
School of Artificial Intelligence and Big Data, Hefei University, Hefei, P. R. China
1 shared publication - Bing Ai
School of Artificial Intelligence and Big Data, Hefei University, Hefei, China
1 shared publication - Chenglong Li
Anhui University
1 shared publication
Latest Publications
- 22 May 2026
- Published by Elsevier in Pattern Recognition
- Vol. 180
Fission-based Dynamic Hypergraph Neural Network
- 22 May 2026
- Published by Elsevier in Pattern Recognition
- Vol. 180
Fission-based Dynamic Hypergraph Neural Network
- 21 June 2025
- Published by World Scientific Pub Co Pte Ltd in International Journal of Wavelets, Multiresolution and Information Processing
Classification ensemble learning via automated graph contrastive learning
In recent years, graph contrastive learning, as a self-supervised learning method, has shown superior performance on unlabeled graph classification tasks and has received extensive attention. However, existing graph contrastive learning methods rely heavily on manual selection of augmentation methods for each dataset, or some empirical methods to accomplish data augmentation. To avoid relying too much on human selection of data augmentation methods, while allowing the model to automatically select negative samples during the training phase, we propose a novel approach called Classification Ensemble Learning via Automated Graph Contrastive Learning (AuCo-CEL). In this paper, the sampling distribution of predefined augmentations is optimized by the Bayesian method to automatically select augmentation schemes. Then, we use a scoring function to evaluate the negative sample difficulty, and automatically select negative samples from easy to hard in each training process. Finally, several random forests are integrated by Adaboost algorithm to complete graph classification tasks. Experiments on seven graph datasets verify that our method outperforms other benchmark models.
- 21 June 2025
- Published by World Scientific Pub Co Pte Ltd in International Journal of Wavelets, Multiresolution and Information Processing
Classification ensemble learning via automated graph contrastive learning
In recent years, graph contrastive learning, as a self-supervised learning method, has shown superior performance on unlabeled graph classification tasks and has received extensive attention. However, existing graph contrastive learning methods rely heavily on manual selection of augmentation methods for each dataset, or some empirical methods to accomplish data augmentation. To avoid relying too much on human selection of data augmentation methods, while allowing the model to automatically select negative samples during the training phase, we propose a novel approach called Classification Ensemble Learning via Automated Graph Contrastive Learning (AuCo-CEL). In this paper, the sampling distribution of predefined augmentations is optimized by the Bayesian method to automatically select augmentation schemes. Then, we use a scoring function to evaluate the negative sample difficulty, and automatically select negative samples from easy to hard in each training process. Finally, several random forests are integrated by Adaboost algorithm to complete graph classification tasks. Experiments on seven graph datasets verify that our method outperforms other benchmark models.
- 29 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 26 (15209210) , 9450-9463
Group Multi-View Transformer for 3D Shape Analysis With Spatial Encoding
In recent years, the results of view-based 3D shape recognition methods have saturated, and models with excellent performance cannot be deployed on memory-limited devices due to their huge size of parameters. To address this problem, we introduce a compression method based on knowledge distillation for this field, which largely reduces the number of parameters while preserving model performance as much as possible. Specifically, to enhance the capabilities of smaller models, we design a high-performing large model called Group Multi-view Vision Transformer (GMViT). In GMViT, the view-level ViT first establishes relationships between view-level features. Additionally, to capture deeper features, we employ the grouping module to enhance view-level features into group-level features. Finally, the group-level ViT aggregates group-level features into complete, well-formed 3D shape descriptors. Notably, in both ViTs, we introduce spatial encoding of camera coordinates as innovative position embeddings. Furthermore, we propose two compressed versions based on GMViT, namely GMViT-simple and GMViT-mini. To enhance the training effectiveness of the small models, we introduce a knowledge distillation method throughout the GMViT process, where the key outputs of each GMViT component serve as distillation targets. Extensive experiments demonstrate the efficacy of the proposed method. The large model GMViT achieves excellent 3D classification and retrieval results on the benchmark datasets ModelNet, ShapeNetCore55, and MCB. The smaller models, GMViT-simple and GMViT-mini, reduce the parameter size by 8 and 17.6 times, respectively, and improve shape recognition speed by 1.5 times on average, while preserving at least 90% of the recognition performance. The code is available at https://github.com/bigdata-graph/GMViT .
- 29 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 26 (15209210) , 9450-9463
Group Multi-View Transformer for 3D Shape Analysis With Spatial Encoding
In recent years, the results of view-based 3D shape recognition methods have saturated, and models with excellent performance cannot be deployed on memory-limited devices due to their huge size of parameters. To address this problem, we introduce a compression method based on knowledge distillation for this field, which largely reduces the number of parameters while preserving model performance as much as possible. Specifically, to enhance the capabilities of smaller models, we design a high-performing large model called Group Multi-view Vision Transformer (GMViT). In GMViT, the view-level ViT first establishes relationships between view-level features. Additionally, to capture deeper features, we employ the grouping module to enhance view-level features into group-level features. Finally, the group-level ViT aggregates group-level features into complete, well-formed 3D shape descriptors. Notably, in both ViTs, we introduce spatial encoding of camera coordinates as innovative position embeddings. Furthermore, we propose two compressed versions based on GMViT, namely GMViT-simple and GMViT-mini. To enhance the training effectiveness of the small models, we introduce a knowledge distillation method throughout the GMViT process, where the key outputs of each GMViT component serve as distillation targets. Extensive experiments demonstrate the efficacy of the proposed method. The large model GMViT achieves excellent 3D classification and retrieval results on the benchmark datasets ModelNet, ShapeNetCore55, and MCB. The smaller models, GMViT-simple and GMViT-mini, reduce the parameter size by 8 and 17.6 times, respectively, and improve shape recognition speed by 1.5 times on average, while preserving at least 90% of the recognition performance. The code is available at https://github.com/bigdata-graph/GMViT .