检索结果-内蒙古大学图书馆

Action recognition using part and attention enhanced feature fusion

Scientific Reports 2025年第1期15卷 1-13页

作者： Danfeng Zhuang Lei Wang Mohamed Sanim Akremi Hedi Tabia Min Jiang School of Automation Wuxi University Wuxi 214105 China IBISC University of Evry Paris-Saclay University 91020 Evry France Jiangsu Provincial Engineering Laboratory of Pattern Recognition and Computational Intelligence Jiangnan University Wuxi 214122 China

In various studies on skeleton-based action recognition, graph convolution has been widely used to extract crucial skeleton information. However, existing methods often face challenges in effectively capturing the motion relationships between distinct body parts, such as the coordination between the torso and limbs. To address these limitations, this paper constructs a novel Part and Attention enhanced Feature Fusion Network (PAFFNet). First, we construct center-of-gravity dynamics within semantic body parts to enhance motion representation and capture dynamic trends. Additionally, we design Joint-based Feature Fusion (JFF) stream and Bone-based Feature Fusion (BFF) stream to extract and fuse complementary information between constructed local parted and global skeleton features for enhanced motion representation. Furthermore, an Improved Adaptive Graph Convolution (IAGC) block with denoising and attention mechanisms is proposed to prioritize critical semantic features. Experiments conducted on the NTU RGB+D and Kinetics-Skeleton datasets demonstrate the effectiveness of PAFFNet in improving skeleton-based action recognition.

关键词： Attention mechanism Feature fusion Multiple streams Part-based structure Skeleton-based action recognition

来源：评论

学校读者我要写书评

暂无评论

Mining High Quality patterns Using Multi-Objective Evolutionary Algorithm (Extended Abstract) 39

Mining High Quality Patterns Using Multi-Objective Evolution...

引用

39th IEEE International Conference on Data engineering, ICDE 2023

作者： Fang, Wei Zhang, Qiang Sun, Jun Wu, Xiaojun Jiangnan University International Joint Laboratory on Artificial Intelligence of Jiangsu Province Jiangsu Provincial Engineering Laboratory of Pattern Recognition and Computational Intelligence Wuxi China

ISBN: (纸本)9798350322279

Most studies on pattern mining have considered only one pattern, such as frequent pattern or high-utility pattern, which is difficult to meet the increasingly diverse needs of users. In this paper, a novel multi-objective problem model for high quality pattern mining (HQPM) is proposed, where the objectives are support, occupancy, and utility. In order to solve the proposed three-objective problem efficiently, an improved multi-objective evolutionary algorithm for HQPM (MOEA-PM) is proposed with two kinds of population initialization strategies and an auxiliary tool. Experimental results on real-world datasets show that the proposed three-objective problem model with the MOEA-PM algorithm can discover patterns that are both frequently occurring and has a high utility in the transaction datasets, while at the same time being relatively complete. The proposed algorithm outperforms the state-of-the-art in terms of efficiency, quality, and convergence speed. © 2023 IEEE.

关键词： Evolutionary algorithms

来源：评论

学校读者我要写书评

暂无评论

CL-Flow:strengthening the normalizing flows by contrastive learning for better anomaly detection

CL-Flow:strengthening the normalizing flows by contrastive l...

引用

2024 International Conference on Image, Signal Processing, and pattern recognition, ISPP 2024

作者： Wang, Shunfeng Li, Yueyang Luo, Haichi Jiangsu Provincial Engineering Laboratory of Pattern Recognition and Computational Intelligence Jiangnan University 1800 Lihu Avenue Jiangsu Wuxi214122 China

ISBN: (纸本)9781510680425

In the anomaly detection field, the scarcity of anomalous samples has directed the current research emphasis towards unsupervised anomaly detection. While these unsupervised anomaly detection methods offer convenience, they also overlook the crucial prior information embedded within anomalous samples. Moreover, among numerous deep learning methods, supervised methods generally exhibit superior performance compared to unsupervised methods. Considering the reasons mentioned above, we propose a self-supervised anomaly detection approach that combines contrastive learning with 2D-Flow to achieve more precise detection outcomes and expedited inference processes. On one hand, we introduce a novel approach to anomaly synthesis, yielding anomalous samples in accordance with authentic industrial scenarios, alongside their surrogate annotations. On the other hand, having obtained a substantial number of anomalous samples, we enhance the 2D-Flow framework by incorporating contrastive learning, leveraging diverse proxy tasks to fine-tune the network. Our approach enables the network to learn more precise mapping relationships from self-generated labels while retaining the lightweight characteristics of the 2D-Flow. Compared to mainstream unsupervised approaches, our selfsupervised method demonstrates superior detection accuracy, fewer additional model parameters, and faster inference speed. Our approach showcases new state-of-the-art results, achieving a performance of 99.6% in image-level AUROC on the MVTecAD dataset and 96.8% in image-level AUROC on the BTAD dataset. © 2024 SPIE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Mutual Information-Guided GA for Bayesian Network Structure Learning (Extended Abstract) 40

Mutual Information-Guided GA for Bayesian Network Structure ...

引用

40th IEEE International Conference on Data engineering, ICDE 2024

作者： Yan, Kefei Fang, Wei Lu, Hengyang Zhang, Xin Sun, Jun Wu, Xiaojun Jiangnan University Jiangsu Provincial Engineering Laboratory of Pattern Recognition and Computational Intelligence Department of Computer Science and Technology Wuxi214122 China

ISBN: (纸本)9798350317152

Bayesian network structure learning (BNSL) from data is an NP-hard problem. Genetic algorithms are powerful for solving combinatorial optimization problems, but the lack of effective guidance results in slow convergence and low accuracy regarding BNSL. To address this problem, we propose a mutual information (MI) guided genetic algorithm (MIGA) for BNSL in this paper, which uses MI to effectively search BN structures. In the initialization phase of MIGA, the population is generated by adding additional constraints based on MI to reach a higher score without losing diversity. By employing normalized MI and defining the population support, the potential dominance in the population can be identified and then used to design a novel crossover operator in order to preserve the dominant genes with a higher probability. Moreover, with the guidance of MI for removing loops from the structures, infeasible solutions can be handled in a straightforward and practical way. The proposed MIGA is evaluated on eleven well-known benchmark datasets and compared with four GA-based methods and four other state-of-the-art BNSL algorithms. Experimental results show that MIGA outperforms the compared algorithms in convergence and learning accuracy. © 2024 IEEE.

关键词： Genetic algorithms

来源：评论

学校读者我要写书评

暂无评论

Advances and applications of machine learning and intelligent optimization algorithms in genome‑scale metabolic network models

引用

Systems Microbiology and Biomanufacturing 2023年第2期3卷 193-206页

作者： Lidan Bai Qi You Chenyang Zhang Jun Sun Long Liu Hengyang Lu Qidong Chen Jiangsu Provincial Engineering Laboratory of Pattern Recognition and Computational Intelligence WuxiJiangsuChina Key Laboratory of Carbohydrate Chemistry and Biotechnology Ministry of EducationJiangnan UniversityWuxi 214122China Science Center for Future Foods Jiangnan UniversityWuxi 214122China

Due to the increasing demand for microbially manufactured products in various industries,it has become important to find optimal designs for microbial cell factories by changing the direction of metabolic flow and its flux size by means of metabolic engineering such as knocking out competing pathways and introducing exogenous pathways to increase the yield of desired ***,with the gradual cross-fertilization between computer science and bioinformatics fields,machine learning and intelligent optimization-based approaches have received much attention in Genome-scale metabolic network models(GSMMs)based on constrained optimization methods,and many high-quality related works have been ***,this paper focuses on the advances and applications of machine learning and intelligent optimization algorithms in metabolic engineering,with special emphasis on ***,the development history of GSMMs is first ***,the analysis methods of GSMMs based on constraint optimization are ***,this paper mainly reviews the development and application of machine learning and intelligent optimization algorithms in genome-scale metabolic *** addition,the research gaps and future research potential in machine learning and intelligent optimization methods applied in GSMMs are discussed.

关键词： Genome-scale metabolic models machine learning Intelligent optimization Metabolic engineering

来源：评论

学校读者我要写书评

暂无评论

STRNet:Triple-stream Spatiotemporal Relation Network for Action recognition

引用

International Journal of Automation and computing 2021年第5期18卷 718-730页

作者： Zhi-Wei Xu Xiao-Jun Wu Josef Kittler School of Artificial Intelligence and Computer Science Jiangnan UniversityWuxi 214122China Jiangsu Provincial Engineering Laboratory of Pattern Recognition and Computational Intelligence Wuxi 214122China Centre for Vision Speech and Signal ProcessingUniversity of SurreyGuildfordGU27XHUK

Learning comprehensive spatiotemporal features is crucial for human action recognition. Existing methods tend to model the spatiotemporal feature blocks in an integrate-separate-integrate form, such as appearance-and-relation network(ARTNet) and spatiotemporal and motion network(STM). However, with blocks stacking up, the rear part of the network has poor interpretability. To avoid this problem, we propose a novel architecture called spatial temporal relation network(STRNet), which can learn explicit information of appearance, motion and especially the temporal relation information. Specifically, our STRNet is constructed by three branches,which separates the features into 1) appearance pathway, to obtain spatial semantics, 2) motion pathway, to reinforce the spatiotemporal feature representation, and 3) relation pathway, to focus on capturing temporal relation details of successive frames and to explore long-term representation dependency. In addition, our STRNet does not just simply merge the multi-branch information, but we apply a flexible and effective strategy to fuse the complementary information from multiple pathways. We evaluate our network on four major action recognition benchmarks: Kinetics-400, UCF-101, HMDB-51, and Something-Something v1, demonstrating that the performance of our STRNet achieves the state-of-the-art result on the UCF-101 and HMDB-51 datasets, as well as a comparable accuracy with the state-of-the-art method on Something-Something v1 and Kinetics-400.

关键词： Action recognition spatiotemporal relation multi-branch fusion long-term representation video classification

来源：评论

学校读者我要写书评

暂无评论

NABP: Social Network Alignment Algorithm based on Pre-trained Language Models 3

NABP: Social Network Alignment Algorithm based on Pre-traine...

引用

3rd International Conference on Artificial intelligence, Automation, and High-Performance Computing, AIAHPC 2023

作者： Ding, Linjun Chen, Jing School of Artificial Intelligence and Computer Science Jiangnan University Jiangsu Wuxi214122 China Jiangsu Provincial Engineering Laboratory of Pattern Recognition and Computing Intelligence Jiangnan University Jiangsu Wuxi214122 China

ISBN: (数字)9781510666580

ISBN: (纸本)9781510666573

The existing social network matching algorithms have problems in processing text attribute information, as they cannot handle polysemy issues of word meanings well and cannot effectively extract deep semantic information from the text. To address the aforementioned issues, this paper introduces pretrained models to extract information from attribute text and proposes a social network alignment algorithm called NABP (Network Alignment based on Pre-trained Language Models). NABP first preprocesses the attribute text data to extract the text feature vectors;then, it uses the DeepWalk algorithm to perform graph representation learning on the two networks to extract their structural features. Next, the attribute feature vectors and structural feature vectors are concatenated to form the final feature vector. Then, a projection function is learned using supervised information to further improve the algorithm's performance. To verify the effectiveness of NABP algorithm, this study conducted experiments on two real social network datasets. The results show that the NABP algorithm performs better than existing algorithms in both the Accuracy and Precision@k metrics, demonstrating the effectiveness of the algorithm and pre-trained language models. © 2023 SPIE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

The optimal graph regularized sparse coding with application to image representation 1

引用

2nd Chinese Conference on pattern recognition and Computer Vision, PRCV 2019

作者： Shu, Zhenqiu Wu, Xiaojun Liu, Zhen You, Congzhe Fan, Honghui School of Computer Engineering Jiangsu University of Technology Changzhou231001 China Jiangsu Provincial Engineering Laboratory of Pattern Recognition and Computational Intelligence Jiangnan University Wuxi214122 China

ISBN: (数字)9783030317232

ISBN: (纸本)9783030317225

Sparse representation has shown its superiority and effectiveness in many real applications in recent years. However, it is still an open problem to effectively preserve the intrinsic geometric structure of data in new representation space. In this paper, we propose a novel method, called the Optimal Graph regularized Sparse Coding (OGSC), to deal with the high dimensional data. Specifically, we impose a rank constraint on the Laplacian matrix of the graph model, and thus can learn the optimal graph to preserve the manifold structure of data in each iteration. Additionally, the optimization scheme for our proposed method is also provided in this paper. The experimental results on three benchmark datasets have shown that our proposed OGSC method outperforms other stat-of-the-art methods. © Springer Nature Switzerland AG 2019.

关键词： Laplace transforms

来源：评论

学校读者我要写书评

暂无评论

Lace Fabric Image Retrieval Using Siamese Neural Network 6

Lace Fabric Image Retrieval Using Siamese Neural Network

引用

6th International Conference on Signal and Image Processing, ICSIP 2021

作者： Xu, Dongdong Li, Yueyang Luo, Haichi Jiangnan University Jiangsu Provincial Engineering Laboratory of Pattern Recognition and Computational Intelligence Wuxi China Jiangnan University College of Internet of Things Engineering Wuxi China

ISBN: (纸本)9780738133737

An efficient lace fabric image retrieval method based on DCNN learning features is proposed in this paper. Fine-tuning with Siamese Neural Network is used to learn effective feature of lace fabric image. During the process of training the Siamese Neural Network, hard negative pairs are selective to achieve fast convergence and good performance. The DCNN learning features are combined with the unique shape feature to enable accurate and efficient retrieval of massive image data. Experimental results demonstrate the effectiveness of retrieval performance of the proposed algorithm and possible practical application of the retrieval system in lace fabric industry to improve management efficiency. © 2021 IEEE.

关键词： Content based retrieval

来源：评论

学校读者我要写书评

暂无评论

Fabric defect detection method based on Cascade Deep Support Vector Data Description 2

Fabric defect detection method based on Cascade Deep Support...

引用

2020 2nd International Conference on Artificial intelligence Technologies and Application, ICAITA 2020

作者： Wang, Xupeng Li, Yueyang Luo, Haichi Jiangsu Provincial Engineering Laboratory of Pattern Recognition and Computational Intelligence Jiangnan University Wuxi China College of Internet of Things Engineering Jiangnan University Wuxi China

The Fabric defect detection method based on Cascade Deep Support Vector Data Description (SVDD) is proposed in this paper. The method describes the data by Deep SVDD to realize the correct evaluation between the normal fabric images and the images with defects in the high dimensional space. Combining with the cascading method, the proposed algorithm can be used to detect minor defects more accurately and quickly. We take the anomaly score as the evaluation result of the testing image, then the Median Absolute Deviation (MAD) outlier detection method is applied to get the final defect detection results. This newly developed method can be trained with only a small amount of defect-free samples, which greatly reduces manual intervention. A variety types of defects in the general fabric images can be detected efficiently. The experimental results demonstrate that the proposed model has good global performance and high recall rate. © 2020 Published under licence by IOP Publishing Ltd.

关键词： Data description

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：