检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Wei, Hongan Liu, Jiaqi Chen, Bo Lin, Liqun Chen, Weiling Zhao, Tiesong The Fujian Key Lab for Intelligent Processing and Wireless Transmission of Media Information College of Physics and Information Engineering Fuzhou University Fujian Fuzhou350116 China Fujian Science & Technology Innovation Laboratory for Optoelectronic Information of China China The Department of Computer Science University of Illinois at Urbana-Champaign UrbanaIL United States

360° videos have received widespread attention due to its realistic and immersive experiences for users. To date, how to accurately model the user perceptions on 360° display is still a challenging issue. In this paper, we exploit the visual characteristics of 360° projection and display and extend the popular just noticeable difference (JND) model to spherical JND (SJND). First, we propose a quantitative 2D-JND model by jointly considering spatial contrast sensitivity, luminance adaptation and texture masking effect. In particular, our model introduces an entropy-based region classification and utilizes different parameters for different types of regions for better modeling performance. Second, we extend our 2D-JND model to SJND by jointly exploiting latitude projection and field of view during 360° display. With this operation, SJND reflects both the characteristics of human vision system and the 360° display. Third, our SJND model is more consistent with user perceptions during subjective test and also shows more tolerance in distortions with fewer bit rates during 360° video compression. To further examine the effectiveness of our SJND model, we embed it in Versatile Video Coding (VVC) compression. Compared with the state-of-the-arts, our SJND-VVC framework significantly reduced the bit rate with negligible loss in visual quality. Copyright © 2023, The Authors. All rights reserved.

关键词： Textures

来源：评论

学校读者我要写书评

暂无评论

Audio-Visual Emotion Recognition Based on a DBN Model with Constrained Asynchrony

Audio-Visual Emotion Recognition Based on a DBN Model with C...

引用

International Conference on Image and Graphics (ICIG)

作者： Danqi Chen Dongmei Jiang Ilse Ravyse Hichem Sahli VUB-NPU Joint Research Group Shaanxi Provincial Key Laboratory on Speech Image and Information Processing Northwestern Polytechnical University Xi'an China Institute of Broadband Technology (IBBT) ETRO Department VUB-NPU Joint Research Group on AVSP Vrije Universiteit Brussel Brussels Belgium

This paper presents an audio visual multi-stream DBN model (Asy_DBN) for emotion recognition with constraint asynchrony, in which audio state and visual state transit individually in their corresponding stream but the transition is constrained by the allowed maximum audio visual asynchrony. Emotion recognition experiments of Asy_DBN with different asynchrony constraints are carried out on an audio visual speech database of four emotions, and compared with the single stream HMM, state synchronous HMM (Syn_HMM) and state synchronous DBN model, as well the state asynchronous DBN model without asynchrony constraint. Results show that by setting the appropriate maximum asynchrony constraint between audio and visual streams, the proposed audio visual asynchronous DBN model gets the highest emotion recognition performance, with an improvement of 15% over Syn_HMM.

关键词： Emotion recognition Hidden Markov models Streaming media Face recognition Speech recognition Humans Bayesian methods Video sequences Face detection Graphics

来源：评论

学校读者我要写书评

暂无评论

MFFLEN: Multi-Label Text Classification Based on Multi-Feature Fusion and Label Embedding

MFFLEN: Multi-Label Text Classification Based on Multi-Featu...

引用

IEEE International Conference on Systems, Man and Cybernetics

作者： Qiliang Gu Shuo Zhao Jianqiang Zhang Gongpeng Song Qin Lu Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center Qilu University of Technology (Shandong Academy of Sciences) Jinan China Faculty of Computer Science and Technology Qilu University of Technology (Shandong Academy of Sciences) Shandong Engineering Research Center of Big Data Applied Technology Jinan China Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China Shandong Branch of China Mobile Communication Group Design Institute Co. Jinan China

ISBN: (数字)9781665410205

ISBN: (纸本)9781665410212

To address the challenges associated with insufficiently extracting and utilizing features at different levels, overlooking the connection between label meanings and text, and facing problems of over-compression or information loss when extracting global information using recurrent neural networks in the field of multi-label text categorization, this paper introduces an innovative model known as MFFLEN (MultiFeature Fusion and Label Embedding Neural Network). First, a back-translated enhanced label set is constructed by back-translated splicing enhancement of the original label set. This set, together with the text, is then input into the embedding layer, which consists of the pre-trained model of bert-baseChinese, thus establishing the initial connection between the text and the labels within the same vector space. Then, to comprehensively extract multi-level semantic features, the model uses a convolutional layer to extract local features and an embedding layer to extract sentence-level features. A bidirectional attention embedded GRU (BAE-GRU) layer is used to extract hybrid finegrained features, which are then fed into the attention layer to further extract hybrid labeled features based on labeling information. Finally, these three different types of features are fused and multi-label text classification results are obtained using a classifier. The experiments proved that the MFFLEN model achieved 73.82% and 88.44% macro-F1 and 88.00% and 88.86% micro-F1 on the two datasets CAIL 2018 Small and CAIL 2018 Split, respectively, which is better than other baseline models.

关键词： Measurement Recurrent neural networks Splicing Text categorization Semantics Multi label classification Logic gates Feature extraction Vectors Data mining

来源：评论

学校读者我要写书评

暂无评论

FireMatch: A Semi-Supervised Video Fire Detection Network Based on Consistency and Distribution Alignment

arXiv

引用

arXiv 2023年

作者： Lin, Qinghua Li, Zuoyong Zeng, Kun Fan, Haoyi Li, Wei Zhou, Xiaoguang School of Computer Science and Mathematics Fujian University of Technology Fuzhou350118 China Fujian Provincial Key Laboratory of Information Processing and Intelligent Control College of Computer and Control Engineering Minjiang University Fuzhou350121 China School of Computer and Artificial Intelligence Zhengzhou University Zhengzhou450001 China School of Automation Beijing University of Post and Telecommunications Beijing100876 China School of Economics and Management Minjiang University Fuzhou350121 China

Deep learning techniques have greatly enhanced the performance of fire detection in videos. However, video-based fire detection models heavily rely on labeled data, and the process of data labeling is particularly costly and time-consuming, especially when dealing with videos. Considering the limited quantity of labeled video data, we propose a semi-supervised fire detection model called FireMatch, which is based on consistency regularization and adversarial distribution alignment. Specifically, we first combine consistency regularization with pseudo-label. For unlabeled data, we design video data augmentation to obtain corresponding weakly augmented and strongly augmented samples. The proposed model predicts weakly augmented samples and retains pseudo-label above a threshold, while training on strongly augmented samples to predict these pseudo-labels for learning more robust feature representations. Secondly, we generate video cross-set augmented samples by adversarial distribution alignment to expand the training data and alleviate the decline in classification performance caused by insufficient labeled data. Finally, we introduce a fairness loss to help the model produce diverse predictions for input samples, thereby addressing the issue of high confidence with the non-fire class in fire classification scenarios. The FireMatch achieved an accuracy of 76.92% and 91.81% on two real-world fire datasets, respectively. The experimental results demonstrate that the proposed method outperforms the current state-of-the-art semi-supervised classification methods. Copyright © 2023, The Authors. All rights reserved.

关键词： Fires

来源：评论

学校读者我要写书评

暂无评论

Convex Dual Theory Analysis of Two-Layer Convolutional Neural Networks with Soft-Thresholding

arXiv

引用

arXiv 2023年

作者： Xiong, Chunyan Lu, Mengli Yu, Xiaotong Cao, Jian Chen, Zhong Guo, Di Qu, Xiaobo The School of Electronic Science and Engineering Fujian Provincial Key Laboratory of Plasma and Magnetic Resonance Biomedical Intelligent Cloud Research and Development Center Xiamen University Xiamen361005 China The School of Computer and Information Engineering University of Technology Xiamen361024 China

Soft-thresholding has been widely used in neural networks. Its basic network structure is a two-layer convolution neural network with soft-thresholding. Due to the network's nature of nonlinearity and nonconvexity, the training process heavily depends on an appropriate initialization of network parameters, resulting in the difficulty of obtaining a globally optimal solution. To address this issue, a convex dual network is designed here. We theoretically analyze the network convexity and numerically confirm that the strong duality holds. This conclusion is further verified in the linear fitting and denoising experiments. This work provides a new way to convexify soft-thresholding neural networks. Copyright © 2023, The Authors. All rights reserved.

关键词： Convex optimization

来源：评论

学校读者我要写书评

暂无评论

Hybridizing infeasibility driven and constrained-domination principle with MOEA/D for constrained multiobjective evolutionary optimization

Lecture Notes in Computer Science (including subseries Lectu...

引用

Lecture Notes in computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 2014年 8916卷 249-261页

作者： Lin, Huibiao Fan, Zhun Cai, Xinye Li, Wenji Wang, Sheng Li, Jian Zhang, Chengdian School of Engineering Shantou University Guangdong515063 China College of Computer Science and Technology Nanjing University of Aeronautics and Astronautics NanjingJiangsu210016 China College of Science Shantou University Guangdong515063 China Guangdong Provincial Key Laboratory of Digital Signal and Image Processing Techniques Shantou University Guangdong515063 China

This paper presents a novel multiobjective constraint handling approach, named as MOEA/D-CDP-ID, to tackle constrained optimization problems. In the proposed method, two mechanisms, namely infeasibility driven (ID) and constrained-domination principle (CDP) are embedded into a prominent multiobjective evolutionary algorithm called MOEA/D. Constraineddomination principle defined a domination relation of two solutions in constraint handling problem. Infeasibility driven preserves a proportion of marginally infeasible solutions to join the searching process to evolve offspring. Such a strategy allows the algorithm to approach the constraint boundary from both the feasible and infeasible side of the search space, thus resulting in gaining a Pareto solution set with better distribution and convergence. The efficiency and effectiveness of the proposed approach are tested on several well-known benchmark test functions. In addition, the proposed MOEA/D-CDP-ID is applied to a real world application, namely design optimization of the two-stage planetary gear transmission system. Experimental results suggest that MOEA/D-CDP-ID can outperform other state-of-the-art algorithms for constrained multiobjective evolutionary optimization. © Springer International Publishing Switzerland 2014.

关键词： Constraint handling

来源：评论

学校读者我要写书评

暂无评论

An ontology-supported inquiry learning technique

Journal of Network Intelligence

引用

Journal of Network Intelligence 2021年第1期6卷 16-28页

作者： Xue, Xingsi Zhang, Jie Yang, Chaofan Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of Technology No.33 Xuefu South Road University Town Minhou Fuzhou Fujian350118 China School of Computer Science and Engineering Yulin Normal University No.299 Education Middle Road Yulin City Guanxi Province537000 China School of Computer Science and Mathematics Intelligent Information Processing Research Center Fujian University of Technology No.33 Xueyuan Road University Town Minhou Fuzhou Fujian350118 China Guangxi Key Laboratory of Automatic Detecting Technology and Instruments Guilin University of Electronic Technology No.1 Jinji Road Guilin Guangxi541004 China

Although Inquiry-based Learning (IBL) offers compelling opportunities for the engineering education, one of the challenges to the implementation of it is that the learner always lacks of the background knowledge. To provide him with the necessary knowledge when making the inquiry, this paper proposes an Ontology Supported Inquiry Learning technique (OSIL), which works on the basis of two ontologies, i.e. curriculum ontology and the learner’s ontology. The experimental results show that the constructed curriculum ontology and learner ontology is of high-quality, and the answers recommended by OSIL almost perfectly match with the true answers. From the experimental results, we can see that it is a feasible way of using the software engineering and prototype evolution method to construct the curriculum and learner ontology, and the individual inquiry based on two ontologies is able to provide the learners with high-quality personalized services. © 2021, Taiwan Ubiquitous information. All rights reserved.

关键词： Curricula

来源：评论

学校读者我要写书评

暂无评论

FERI: Feature Enhancement and Relational Interaction for Image-text Matching

FERI: Feature Enhancement and Relational Interaction for Ima...

引用

International Conference on Parallel and Distributed Systems (ICPADS)

作者： Yu Zhang Jianqiang Zhang Gongpeng Song Qin Lu Shuo Zhao Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Engineering Research Center of Big Data Applied Technology Faculty of Computer Science and Technology Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China Shandong Branch of China Mobile Communication Group Design Institute Co. Jinan China

ISBN: (数字)9798331515966

ISBN: (纸本)9798331515973

Image-text matching is an important problem at the intersection of computer vision and natural language processing. It aims to establish the semantic link between image and text to achieve high-quality semantic alignment between the two modalities. However, the existing methods have the problem that the meaning expressed in the image or the complex narrative in the text cannot be fully understood due to insufficient feature extraction. Moreover, due to the essential modal differences between images and texts, how to effectively and accurately align the semantic contents in images and texts has become the key of research. In order to solve the above problems, this paper proposes a method based on feature enhancement and relationship interaction. When processing images, the proposed method fuses labeled features, region features and location features to represent images. When processing text, a combination of Bi-GRU and self-attention mechanism is used to represent the text. In order to further align the semantic content in images and texts accurately, this paper improves two relational interaction mechanisms by identifying connection relationships and learning association relationships. Thus, the relation enhanced embedding is obtained. Finally, it calculated the similarity of the enhanced embedding to judge the matching degree of the image and text. Extensive experiments on the public datasets Flickr30K and MSCOCO demonstrate the effectiveness of our method.

关键词： computer vision Fuses Image matching Semantics Feature extraction Natural language processing

来源：评论

学校读者我要写书评

暂无评论

Pixel-Level Polyp Segmentation Network Based on Parallel Feature Enhancement and Attention Mechanism

SSRN

引用

SSRN 2024年

作者： Xu, Haiping Wang, Jie Li, Zuoyong Cheng, Xuesong Teng, Shenghua College of Computer and Data Science Minjiang University Fuzhou China School of Information and Communication Engineering University of Electronic Science and Technology of China Chengdu China College of Electronic and Information Engineering Shandong University of Science and Technology Qingdao China Fujian Key Laboratory of Medical Big Data Engineering Fujian Provincial Hospital Fuzhou China

Polyp segmentation, a crucial tool in medical diagnostics, effectively locates and segments polyps, thereby aiding clinicians in enhancing both detection efficiency and accuracy. To address the persistent challenges of imprecise polyp localization and ambiguous boundary segmentation, we propose a pixel-level polyp segmentation network based on a parallel feature enhancement and attention mechanism (PFEA-Net). The proposed PFEA-Net follows an enconder-decoder architecture. Given that polyps vary greatly in size and shape, PFEA-Net first utilizes the HarDNet68 as the encoder to extract multi-level feature representations. Then, a parallel feature enhancement module is designed to enrich the feature map of each level by integrating semantic information derived from higher-level features. Meanwhile, a parallel partial decoder module is used to introduce global contextual information. Finally, the reverse attention module refines the boundary features. PFEA-Net effectively extracts high-level representations to enhance feature depiction, capturing both detailed and semantic information in the images. Rigorous experiments conducted on five publicly available polyp datasets using six evaluation metrics conclusively demonstrate that the proposed network exhibits superior robustness and achieves more precise segmentation. Compared to other state-of-the-art algorithms, PFEA-Net distinguishes itself with exceptional performance. © 2024, The Authors. All rights reserved.

关键词： Pixels

来源：评论

学校读者我要写书评

暂无评论

A Prediction Method for Atmospheric Monitoring Data Based on Transformer and GAN

A Prediction Method for Atmospheric Monitoring Data Based on...

引用

IEEE information technology, Networking, Electronic and Automation Control Conference

作者： Hongge Zhao Jingli Wang Tixiang Zhang Huijuan Hao Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China Jinan Eco-environmental Bureau Pingyin Branch Office Jinan China Shandong Unite Environmental Technology Co Jinan China

ISBN: (数字)9798350370805

ISBN: (纸本)9798350370812

With the rapid development of urbanization and industrialization, atmospheric pollution has become a major environmental problem of global concern. The problem of atmospheric pollution not only destroys the ecological en-vironment of the earth, but also has a serious impact on people's health and development. Therefore, accurate prediction of atmospheric monitoring data is of great significance to environmental management as well as public health. To improve the accuracy of prediction, we propose a prediction method for atmospheric monitoring data based on Trans-former and Generative Adversarial Network (GAN). Firstly, data pre-processing is performed on the collected atmospheric monitoring data, and then, constructing a prediction model based on Transformer and GAN for atmospheric monitoring data, aiming at extracting the data features of atmospheric monitoring data by Transformer and generating the data by using the generator of GAN. In addition, we improve the position encoding (PE) of the Transformer in order to it can efficiently extract temporal features of atmospheric monitoring data. Finally, the effectiveness of the proposed method is verified by data from three air monitoring stations in a city.

关键词： Accuracy Pollution Atmospheric modeling Urban areas Predictive models Generative adversarial networks Transformers Feature extraction Data models Monitoring

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：