检索结果-内蒙古大学图书馆

A Survey of Human Action Recognition and Posture Prediction

Tsinghua Science and technology 2022年第6期27卷 973-1001页

作者： Nan Ma Zhixuan Wu Yiu-ming Cheung Yuchen Guo Yue Gao Jiahong Li Beiyan Jiang Beijing Key Laboratory of Information Service Engineering the College of RoboticsBeijing Union UniversityBeijing 100101China Department of Computer Science Hong Kong Baptist UniversityHong Kong 999077China Beijing National Research Center for Information Science and Technology Tsinghua UniversityBeijing 100084China School of Software Tsinghua UniversityBeijing 100084China College of Robotics Beijing Union UniversityBeijing 100101China

Human action recognition and posture prediction aim to recognize and predict respectively the action and postures of persons in *** are both active research topics in computer vision community,which have attracted considerable attention from academia and *** are also the precondition for intelligent interaction and human-computer cooperation,and they help the machine perceive the external *** the past decade,tremendous progress has been made in the field,especially after the emergence of deep learning ***,it is necessary to make a comprehensive review of recent *** this paper,firstly,we attempt to present the background,and then discuss research ***,we introduce datasets,various typical feature representation methods,and explore advanced human action recognition and posture prediction ***,facing the challenges in the field,this paper puts forward the research focus,and introduces the importance of action recognition and posture prediction by taking interactive cognition in self-driving vehicle as an example.

关键词： human action recognition posture prediction computer vision human-computer cooperation interactive cognition

来源：评论

学校读者我要写书评

暂无评论

An End-to-End Graph Attention Network Hashing for Cross-Modal Retrieval 38

An End-to-End Graph Attention Network Hashing for Cross-Moda...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Jin, Huilong Zhang, Yingxue Shi, Lei Zhang, Shuang Kou, Feifei Yang, Jiapeng Zhu, Chuangying Luo, Jia College of Engineering Hebei Normal University China College of Computer and Cyber Security Hebei Normal University China State Key Laboratory of Media Convergence and Communication Communication University of China China Key Laboratory of Education Informatization for Nationalities Yunnan Normal University Ministry of Education China School of Computer Science Beijing University of Posts and Telecommunications China Guangxi Key Laboratory of Trusted Software Guilin University of Electronic Technology China Chongqing Research Institute Beijing University of Technology China

Due to its low storage cost and fast search speed, cross-modal retrieval based on hashing has attracted widespread attention and is widely used in real-world applications of social media search. However, most existing hashing methods are often limited by uncomprehensive feature representations and semantic associations, which greatly restricts their performance and applicability in practical applications. To deal with this challenge, in this paper, we propose an end-to-end graph attention network hashing (EGATH) for cross-modal retrieval, which can not only capture direct semantic associations between images and texts but also match semantic content between different modalities. We adopt the contrastive language image pretraining (CLIP) combined with the Transformer to improve understanding and generalization ability in semantic consistency across different data modalities. The classifier based on graph attention network is applied to obtain predicted labels to enhance cross-modal feature representation. We construct hash codes using an optimization strategy and loss function to preserve the semantic information and compactness of the hash code. Comprehensive experiments on the NUS-WIDE, MIRFlickr25K, and MS-COCO benchmark datasets show that our EGATH significantly outperforms against several state-of-the-art methods. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Disentangled counterfactual learning for physical audiovisual commonsense reasoning 23

Disentangled counterfactual learning for physical audiovisua...

引用

Proceedings of the 37th International Conference on Neural Information Processing Systems

作者： Changsheng Lv Shuai Zhang Yapeng Tian Mengshi Qi Huadong Ma Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia and Beijing University of Posts and Telecommunications Department of Computer Science The University of Texas at Dallas

In this paper, we propose a Disentangled Counterfactual Learning (DCL) approach for physical audiovisual commonsense reasoning. The task aims to infer objects' physics commonsense based on both video and audio input, with the main challenge is how to imitate the reasoning ability of humans. Most of the current methods fail to take full advantage of different characteristics in multi-modal data, and lacking causal reasoning ability in models impedes the progress of implicit physical knowledge inferring. To address these issues, our proposed DCL method decouples videos into static (time-invariant) and dynamic (time-varying) factors in the latent space by the disentangled sequential encoder, which adopts a variational autoencoder (VAE) to maximize the mutual information with a contrastive loss function. Furthermore, we introduce a counterfactual learning module to augment the model's reasoning ability by modeling physical knowledge relationships among different objects under counterfactual intervention. Our proposed method is a plug-and-play module that can be incorporated into any baseline. In experiments, we show that our proposed method improves baseline methods and achieves state-of-the-art performance. Our source code is available at https://***/Andy20178/DCL.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Expression Recognition and intelligent Classroom State Mining in Teaching Videos Based on Semi-Supervised Learning Generative Adversarial Network

Expression Recognition and Intelligent Classroom State Minin...

引用

Data-driven Optimization of Complex Systems (DOCS), International Conference on

作者： Ruoyu Fan Meiyu Liang Mengran Yin Junping Du Beijing Key Laboratory of Intelligent Communication Software Multimedia School of Computer Science (National Pilot Software Engineering School) Beijing University of Posts and Telecommunications Beijing China

We constructed a student classroom teaching video dataset, constructed a student classroom expression dataset through video frame extraction, face detection, face alignment, and facial extraction, and constructed a student classroom behavior dataset through object detection algorithm. Aiming at the characteristics that the annotation information of the students' classroom expression dataset constructed in this paper is limited and the data of some categories are less, an expression recognition method based on semi-supervised learning generative adversarial network (SSLGAN) is proposed. The problem of limited annotation information is solved through semi-supervised learning, and the problem of less data of some categories is solved through data enhancement of generative adversarial network. An intelligent classroom state mining method that integrates student expressions and behaviors has been proposed, achieving intelligent classroom state mining of teaching videos and automatic evaluation of classroom listening status for individual and overall students in different types of classrooms, such as interactive classes, discussion classes, and self-study classes. Compared with CNN, RMN, Temporal Ensembling and Mean Teacher, the proposed SSLGAN has achieved the highest accuracy in the benchmark dataset FER2013, JAFFE, RaFD and self-built student classroom expression dataset. In the case where there are few labels in the self-built student classroom expression dataset, the accuracy of student expression recognition results still reaches 76%, and the proposed intelligent classroom state mining method that integrates student expressions and behaviors for different types of classrooms also has high accuracy on the self-built student classroom teaching video dataset.

关键词：

来源：评论

学校读者我要写书评

暂无评论

DGNN: Decoupled Graph Neural Networks With Structural Consistency Between Attribute and Graph Embedding Representations

引用

IEEE Transactions on Big Data 2024年

作者： Wang, Jinlu Guo, Jipeng Sun, Yanfeng Gao, Junbin Wang, Shaofan Yang, Yachao Yin, Baocai Beijing University of Technology Beijing Key Laboratory of Multimedia and Intelligent Software Technology Faculty of Information Technology Beijing100124 China Beijing University of Chemical Technology College of Information Science and Technology Beijing100029 China The University of Sydney Business School The University of Sydney Discipline of Business Analytics CamperdownNSW2006 Australia

Graph neural networks (GNNs) exhibit a robust capability for representation learning on graphs with complex structures, demonstrating superior performance across various applications. Most existing GNNs utilize graph convolution operations that integrate both attribute and structural information through coupled way. And these GNNs, from an optimization perspective, seek to learn a consensus and compromised embedding representation that balances attribute and graph information, selectively exploring and retaining valid information in essence. To obtain a more comprehensive embedding representation, a novel GNN framework, dubbed Decoupled Graph Neural Networks (DGNN), is introduced. DGNN separately explores distinctive embedding representations from the attribute and graph spaces by decoupled terms. Considering that the semantic graph, derived from attribute feature space, contains different node connection information and provides enhancement for the topological graph, both topological and semantic graphs are integrated by DGNN for powerful embedding representation learning. Further, structural consistency between the attribute embedding and the graph embedding is promoted to effectively eliminate redundant information and establish soft connection. This process involves facilitating factor sharing for adjacency matrices reconstruction, which aims at exploring consensus and high-level correlations. Finally, a more powerful and comprehensive representation is achieved through the concatenation of these embeddings. Experimental results conducted on several graph benchmark datasets demonstrate its superiority in node classification tasks. © 2015 IEEE.

关键词： Graph embeddings

来源：评论

学校读者我要写书评

暂无评论

Federal Learning Based COVID-19 Fake News Detection With Deep Self-Attention Network 7

Federal Learning Based COVID-19 Fake News Detection With Dee...

引用

7th IEEE International Conference on Cloud Computing and Intelligence Systems, CCIS 2021

作者： Ouyang, Suyu Du, Junping Wang, Benzhi Yu, Bowen Wang, Yuhui Liang, Meiyu Beijing University of Posts and Telecommunications Beijing Key Laboratory of Intelligent Telecommunication Software and Multimedia School of Computer Science Beijing100876 China

ISBN: (纸本)9781665441490

As social media becomes more and more popular, fake news spreads rapidly which is more likely to cause serious consequences, especially during the COVID-19 pandemic. On the premise of meeting data privacy and security requirements, federated learning uses multi-party heterogeneous data to further promote machine learning. This paper proposes a federal learning based COVID-19 fake news detection model with deep self-attention network (FL-FNDM). We construct a deep self-attention network for fake news detection, which combines self-attention-based pretrained model BERT and deep convolutional neural network to detect fake news. Moreover, the fake news detection model is learned under the framework of horizontal federated learning, aiming at protecting users' data security and privacy. The experimental results demonstrate that the proposed model can improve the performance of fake news detection on the COVID-19 dataset, which can achieve almost the same effect of sharing data without leaking user data. © 2021 IEEE.

关键词： COVID-19

来源：评论

学校读者我要写书评

暂无评论

Federated Learning Method Based on Knowledge Distillation and Deep Gradient Compression 7

Federated Learning Method Based on Knowledge Distillation an...

引用

7th IEEE International Conference on Cloud Computing and Intelligence Systems, CCIS 2021

作者： Cui, Haiyan Du, Junping Jiang, Yang Wang, Yue Yu, Runyu Beijing University of Posts and Telecommunications Beijing Key Laboratory of Intelligent Telecommunication Software and Multimedia School of Computer Science Beijing100876 China

ISBN: (纸本)9781665441490

Federated learning is a new type of multi-agency collaborative training model paradigm, which is widely used in many fields, among which communication overhead is a key issue. In order to reduce the amount of data transmitted in the communication process, we propose a federated learning algorithm based on knowledge distillation and deep gradient compression (Fed-KDDGC-SGD). First, we use local data on the client to train the teacher network, and then use the soft labels generated by the teacher network to train the student network and upload the gradient to the central server during the training process. In order to further reduce the communication bandwidth occupied by sending the gradient, the deep gradient compression algorithm is used to compress the gradient vector, and only the gradient value of the top R% of the absolute value is sent. The experimental results show that the improved federated learning algorithm effectively reduces the communication overhead and has certain practical significance. © 2021 IEEE.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

A Hierarchical Multi-label Classification Algorithm for Scientific Papers Based on Graph Attention Networks 1st

A Hierarchical Multi-label Classification Algorithm for Scie...

引用

1st CAAI International Conference on Artificial Intelligence, CICAI 2021

作者： Zheng, Changwei Xue, Zhe Du, Junping Kou, Feifei Liang, Meiyu Xu, Mingying Beijing Key Laboratory of Intelligent Telecommunication Software and Multimedia School of Computer Science Beijing University of Posts and Telecommunications Beijing100876 China

ISBN: (纸本)9783030930455

Scientific paper classification refers to assigning one or more subject categories to papers. This task requires a lot of domain knowledge and heavy manual annotation. With the gradual increase in interdisciplinary research, a paper often has multiple categories. For instance, both Chinese Library Classification (http://***/ ) and Engineering Village (EI) have a complete classification system, and there is a hierarchical relationship between the categories. The category of the paper has a hierarchical structure, so the paper classification can be converted into a hierarchical classification problem. However, the existing methods cannot effectively classify papers due to the following two reasons: First, these methods cannot well capture the semantic relationship between papers. Second, they neglect to model the hierarchical structure of labels. In this paper, we propose a hierarchical label attention model based on graph attention network, which utilizes word co-occurrence to model the semantic relationship of papers. We use multiple linear layers to model the category hierarchy and combine every hierarchy of labels through an attention mechanism. The experiments are conducted on CNKI (https://***/ ) and RCV1 datasets. The experimental results demonstrate that our method is superior to the other methods in the task of scientific paper classification. © 2021, Springer Nature Switzerland AG.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

Multi-view Relevance Matching Model of Scientific Papers Based on Graph Convolutional Network and Attention Mechanism 1st

Multi-view Relevance Matching Model of Scientific Papers Bas...

引用

1st CAAI International Conference on Artificial Intelligence, CICAI 2021

作者： Song, Jie Xue, Zhe Du, Junping Kou, Feifei Liang, Meiyu Xu, Mingying Beijing Key Laboratory of Intelligent Telecommunication Software and Multimedia School of Computer Science Beijing University of Posts and Telecommunications Beijing100876 China

ISBN: (纸本)9783030930455

Deep learning has been widely used in text matching tasks. However, the existing deep learning models are mainly designed for short texts matching and cannot be directly applied to the search of scientific papers. The main reason is that the differences between long and short texts in scientific paper search have not been fully considered, and the structural information of the text will be lost when the length difference is large. In order to solve the above long-short scientific text matching problem, we propose a multi-view relevance matching model (MVRM) of scientific papers based on graph convolutional network and attention mechanism. First, we use scientific papers abstract to construct interactive graph to retain the structural information in the long text. Each vertex denotes the keyword in the abstract and the edge weight denotes similarity between the keywords. Second, we propose a matching network for interactive graph based on the graph convolution networks. Multiple keywords in the search term form multiple views, and each keyword under each view interacts with the interactive graph. Then the interaction feature vectors from multiple views are generated through graph convolution network. Finally, attention mechanism is used for fusion, and the final matching result is output through the multilayer perceptron. Experiments on several representative scientific paper search datasets demonstrate that our model achieves better performance. © 2021, Springer Nature Switzerland AG.

关键词： Graphic methods

来源：评论

学校读者我要写书评

暂无评论

Financial Topic Detection Algorithm Based on Multi-feature Fusion

Financial Topic Detection Algorithm Based on Multi-feature F...

引用

Chinese intelligent Systems Conference, CISC 2020

作者： Zhang, Qiang Du, Junping Kou, Feifei Beijing Key Laboratory of Intelligent Telecommunication Software and Multimedia School of Computer Science Beijing University of Posts and Telecommunications Beijing100876 China

ISBN: (纸本)9789811584572

In order to help investors get effective information from a large amount of financial news data as soon as possible, a topic detection model based on multi-view text semantics and clustering topic detection algorithm is proposed. In the financial news data set, different models are used to extract the characteristics of the news, and the characteristics of various models are merged. The clustering algorithm is improved by introducing JS divergence and time decay factors. The experimental results show that compared with the traditional topic detection model, the proposed method has higher accuracy of topic detection and shorter runtime of clustering algorithm. © 2021, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Clustering algorithms

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：