检索结果-内蒙古大学图书馆

image processing, Computer Vision and Machine Learning (ICICML), International Conference on

作者： Yuhao Su Jianliang Zhang Hua Liu Jing Zhang La Consolacion University Philippines Philippines Taishan Polytechnic TaiAn China Henan Agricultural University HeNan China

ISBN: (数字)9798350355413

ISBN: (纸本)9798350355420

In the context of educational informatization, the rapid growth of educational big data has provided new data support for classroom management and educational decision-making. However, anomaly detection in this data still faces both technical and theoretical challenges. This paper explores the anomaly detection methods in educational big data based on deep neural networks from a theoretical perspective, analyzing the characteristics of educational data and the difficulties associated with detecting anomalies. By combining traditional anomaly detection models with deep learning methods, this paper outlines the theoretical advantages of deep neural networks in anomaly detection and discusses their potential applications in educational management. Finally, the paper proposes theoretical directions for further optimizing deep neural network models in the education sector, offering support for classroom management and educational optimization.

关键词： Deep learning Adaptation models Decision making Transfer learning artificial neural networks Big Data Data models General Data Protection Regulation Anomaly detection Optimization

来源：评论

学校读者我要写书评

暂无评论

PBGAN: Path Based Graph Attention Network for Heterophily 26

PBGAN: Path Based Graph Attention Network for Heterophily

引用

26th International Conference on Pattern Recognition / 8th International Workshop on image Mining - Theory and applications (IMTA)

作者： Guan, Xiangshuo Wang, Dan Xiong, Congcong Li, Shangrong Chen, Youning Tianjin Univ Sci & Technol Coll Artificial Intelligence Tianjin 300457 Peoples R China

ISBN: (数字)9781665490627

ISBN: (纸本)9781665490627

Typical Graph neural networks (GNNs) have achieved noticeable performance in node classification tasks with the strategy of neighbor aggregation, which benefits from the homophily assumption of graphs (nodes with the same class tend to be connected). However, the performance of many models degrades tensely when data tend to be heterophilic. Therefore, we need to optimize the neighbor aggregation strategy to deal with these two different types of graphs. In this paper, we propose a Path Based Graph Attention Network (PBGAN) to address this issue. It utilizes structure and path information in the high-order neighborhood and employs Recurrent neural Network (RNN) to learn attention coefficients according to the shortest paths between node pairs. This learning method makes the attention coefficients better reflect the relational importance between nodes, which is practical for processing heterophilic graphs. Empirical results demonstrate that the PBGAN achieves comparable performance in both homophilic and heterophilic datasets, proving the effectiveness of the proposed attention learning method.

关键词： Graph neural networks Heterophilic Graph Attention Mechanism Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

Sign Language Recognition using Spiking neural networks

引用

Procedia Computer Science 2024年 235卷 2674-2683页

作者： Pranav Chaudhari Alex Vicente-Sola Amlan Basu Davide L. Manna Paul Kirkland Gaetano Di Caterina Department of Information and Communication Technology Manipal Institute of Technology Manipal Academy of Higher Education Manipal 576104 India Neuromorphic Sensor Signal Processing Lab Centre for Image and Signal Processing Electrical and Electronic Engineering Department University of Strathclyde Glasgow UK

In recent years, research in automatic Sign Language Recognition (SLR) has undergone significant progress, serving as a founda-tional base for developing applications that aim to promote the integration of deaf individuals into society. Most of this progress is owed to the recent developments in deep learning. However, the deployment of conventional artificial neural networks (ANNs) can be hindered by their requirements in terms of computational power and energy consumption. Therefore, to improve the ef-ciency of current SLR systems, in this work, we propose the use of the increasingly popular Spiking neural networks (SNNs), which, on the one hand, provide more energy-efficient computations than conventional ANNs and, on the other hand, are able to process temporal sequences with simpler architectures thanks to their temporal dynamics. To evaluate our method, we utilize WLASL300, the 300-word (300 classes of signs) dataset from Word-Level American Sign Language, and achieve an improvement in accuracy with the SNN (+2.70%) over the previous state-of-the-art, when working with energy-efficient spiking neurons. Furthermore, we construct a non-spiking version of the same network and evaluate it in a similar manner. Our results demonstrate how the SNN has sparser activations (25% less), thanks to the use of spiking neurons, and therefore can be implemented with a lower power requirement than an ANN version of the same architecture. This work thus demonstrates the possibility of performing SLR in a very effective and efficient way, thus opening up the development of applications that span from the automatic real-time translation of dynamic signs to remote control utilizing sign languages.

关键词： Spiking neural networks American Sign Language Sign Language Recognition Deep Learning

来源：评论

学校读者我要写书评

暂无评论

Topological Insulator Optoelectronic Synapses for High-Accuracy Binary image Recognition using Recurrent neural networks

引用

ADVANCED OPTICAL MATERIALS 2023年第2期11卷

作者： Wan, Zhengfen Zhang, Qiwen Hu, Fangzhen Dong, Yibo Li, Runze Hu, Liangchen Xie, Yiyang Yue, Zengji Chen, Xi Gu, Min Univ Shanghai Sci & Technol Inst Photon Chips Shanghai 200093 Peoples R China Univ Shanghai Sci & Technol Ctr Artificial Intelligence Nanophoton Sch Opt Elect & Comp Engn Shanghai 200093 Peoples R China Beijing Univ Technol Key Lab Optoelect Technol Minist Educ Beijing 100124 Peoples R China

artificial optoelectronic synapses have drawn tremendous attention in neuromorphic computing due to their exceptional properties of incorporating optical-sensing and synaptic functions. However, the complex fabrication processes and device architectures greatly limit their applications. More importantly, artificial neural networks (ANNs) commonly implemented with optoelectronic synapses cannot take full advantage of the time-dependent data of synaptic devices, resulting in defective accuracies. Here, facile two-terminal optoelectronic synapses based on topological insulator Sb2Te3 films are fabricated, which exhibit significant photocurrent responses, owing to the efficient light-matter interaction in bulk and the topological surface state of Sb2Te3. The performance of Sb2Te3 devices can be tuned both optically and electrically. Typical characteristics of synapses, such as paired-pulse facilitation, short-term memory, long-term memory, and learning behavior, have been demonstrated. With the establishment of recurrent neural networks (RNNs) that are committed to processing temporal data, the as-fabricated synapse devices are employed for binary image recognition of handwritten numbers "0" and "1". The recognition accuracy of RNNs can reach as high as 100%, which is dramatically higher than those of ANNs. The effective employment of temporal data with RNNs ensured high recognition accuracy. These Sb2Te3 optoelectronic synapses with RNNs indicate the great potential for developing high-performance brain-inspired neuromorphic computing.

关键词： image recognition neuromorphic computing optoelectronic synapses recurrent neural networks Sb Te-2 (3) films topological insulators

来源：评论

学校读者我要写书评

暂无评论

Sparse Auto Encoder Based Deep neural Network for image processing Technology

Sparse Auto Encoder Based Deep Neural Network for Image Proc...

引用

Network, Multimedia and Information Technology (NMITCON) International Conference on

作者： Zufang Yang Heqiong Zeng College of Artificial Intelligence Wuhan Technology and Business University Wuhan China Institute of Information and Intelligent Engineering Applications Wuhan Technology and Business University Wuhan China

ISBN: (数字)9798350372892

ISBN: (纸本)9798350372908

The image processing technology has been widely used in various field including video and image surveillance automatic capture the various application in image way representing. However, image process in different way and maximum amount of linear layer then easily fall into local optimum and occur overfitting in learn the data. This paper proposed Sparse Auto Encoder based Deep neural Network (SAE-DNN) for image classification is efficiently achieve high accuracy especially noisy data and dimensionality reduces. The SAE-DNN technique help to remove noise in data by learning the image pattern efficiently, which improve classification accuracy in noisy environments. Initial data is obtained from CIFAR-100 dataset and Pre-processing involves data augmentation, where transformations like translation and rotation are frequently applied during training phase of neural networks for image classification. The MobileNetV2 involved feature extraction, which efficiently extracts relevant features from training data. The proposed SAE-DNN technique are evaluated achieve better outcomes such as accuracy of 92.02% on CIFAR100 dataset existing technique like Fully Convolutional neural Network (FCNN) and Mobile Net V2 with Wavelet-Attention Convolutional neural Network (WA-CNN) are evaluated for proposed method.

关键词： Training Accuracy Noise Training data Feature extraction Data augmentation Noise measurement Convolutional neural networks Data mining image classification

来源：评论

学校读者我要写书评

暂无评论

Blind quality-based pairwise ranking of contrast changed color images using deep networks

引用

SIGNAL processing-image COMMUNICATION 2024年 121卷

作者： Chetouani, Aladine Qureshi, Muhammad Ali Deriche, Mohamed Beghdadi, Azeddine Univ Orleans Lab PRISME Orleans France Islamia Univ Bahawalpur Bahawalpur 63100 Pakistan Ajman Univ Coll Engn Artificial Intelligence Res Ctr Ajman U Arab Emirates Univ Sorbonne Paris Nord Inst Galilee L2TI Villetaneuse France

Next-generation multimedia networks are expected to provide systems and applications with top Quality of Experience (QoE) to users. To this end, robust quality evaluation metrics are critical. Unfortunately, most current research focuses only on modeling and evaluating mainly distortions across the pipeline of multimedia networks. While distortions are important, it is also as important to consider the effects of enhancement and other manipulations of multimedia content, especially images and videos. In contrast to most existing works dedicated to evaluating image/video quality in its traditional context, very few research efforts have been devoted to image Quality Enhancement Assessment (IQEA) and more specifically, Contrast Enhancement Evaluation (CEE). Our contribution fills this gap by proposing a pairwise ranking scheme for estimating and evaluating the perceptual quality of image contrast change (contrast enhancement and/or contrast-distorted images) process. We propose a novel Deep Learning-based Blind Quality pairwise Ranking scheme for Contrast-Changed (Deep-BQRCC) images. This method provides an automatic pairwise ranking of a set of contrast-changed images. The proposed framework is based on using a pair of Convolutional neural networks (CNN) together with a saliency-based attention model and a color-difference visual map. Extensive experiments were conducted to validate the effectiveness of the proposed workflow through an ablation analysis. Different combinations of CNN models and pooling strategies were analyzed. The proposed Deep-BQRCC approach was evaluated over three dedicated publicly available datasets. The experimental results showed an increase in performance within a range of 3-10% compared to state-of-the-art IQEA measures.

关键词： Contrast enhancement evaluation (CEE) Contrast change Contrast enhancement (CE) image quality enhancement evaluation (IQEA) Convolutional neural network (CNN) Visual saliency image ranking

来源：评论

学校读者我要写书评

暂无评论

Non-symmetric over-time pooling using pseudo-grouping functions for convolutional neural networks

引用

ENGINEERING applications OF artificial INTELLIGENCE 2024年第PartE期133卷

作者： Ferrero-Jaurrieta, Mikel Paiva, Rui Cruz, Anderson Bedregal, Benjamin De Miguel, Laura Takac, Zdenko Lopez-Molina, Carlos Bustince, Humberto Univ Publ Navarra Dept Stat Comp Sci & Math Pamplona Spain Inst Fed Educ Ciencia & Tecnol Ceara Fortaleza Brazil Univ Fed Rio Grande Do Norte Inst Metropole Digital Natal Brazil Univ Fed Rio Grande Do Norte Dept Informat & Matemat Aplicada Natal Brazil Slovak Univ Technol Fac Mat Sci & Technol Trnava Slovakia Univ Publ Navarra UPNA IdiSNA Navarrabiomed Hosp Univ Navarra HUN Pamplona 31008 Spain

Convolutional neural networks (CNNs) are a family of networks that have become state-of-the-art in several fields of artificial intelligence due to their ability to extract spatial features. In the context of natural language processing, they can be used to build text classification models based on textual features between words. These networks fuse local features to generate global features in their over-time pooling layers. These layers have been traditionally built using the maximum function or other symmetric functions such as the arithmetic mean. It is important to note that the order of input local features is significant (i.e. the symmetry is not an inherent characteristic of the model). While this characteristic is appropriate for image-oriented CNNs, where symmetry might make the network robust to image rigid transformations, it seems counter-productive for text processing, where the order of the words is certainly important. Our proposal is, hence, to use non-symmetric pooling operators to replace the maximum or average functions. Specifically, we propose to perform over-time pooling using pseudo-grouping functions, a family of non-symmetric aggregation operators that generalize the maximum function. We present a construction method for pseudo-grouping functions and apply different examples of this family to over-time pooling layers in text-oriented CNNs. Our proposal is tested on seven different models and six different datasets in the context of engineering applications, e.g. text classification. The results show an overall improvement of the models when using non-symmetric pseudo-grouping functions over the traditional pooling function.

关键词： Pseudo-grouping function Over-time pooling Text classification Feature fusion Aggregation function

来源：评论

学校读者我要写书评

暂无评论

Research on image Recognition Technology Based on Multimodal Deep Learning

Research on Image Recognition Technology Based on Multimodal...

引用

image processing and Computer applications (ICIPCA), IEEE International Conference on

作者： Jinyin Wang Xingchen Li Yixuan Jin Yihao Zhong Keke Zhang Chang Zhou Stony Brook University New York USA University of Southern California Los Angeles USA Independent Researchers New York USA New York University New York USA Northeastern University Massachusetts USA Columbia University New York USA

ISBN: (数字)9798350360240

ISBN: (纸本)9798350384161

This research explores a human multi-modal behavior identification algorithm using deep neural networks. The algorithm leverages various deep neural networks tailored to different types of modal information to process diverse video data. By integrating these neural networks, the algorithm effectively identifies behaviors across multiple modalities. Data collection for this study utilized multiple cameras developed by Microsoft Kinect, capturing skeletal point data in addition to conventional images. This dual data collection method enables the extraction of motion features from the images. The synthesized behavioral characteristics from these data sources facilitate precise behavior identification and categorization. The proposed algorithm's effectiveness was validated using the MSR3D dataset. Experimental results demonstrate that behavior recognition accuracy remains consistent across various scenarios, indicating the robustness of the recognition process. The findings reveal that the algorithm substantially enhances the accuracy of pedestrian behavior recognition in video footage.

关键词： Deep learning Technological innovation image recognition Accuracy Soft sensors artificial neural networks Data collection Feature extraction Robustness Behavioral sciences

来源：评论

学校读者我要写书评

暂无评论

A Comprehensive Survey on Graph neural networks

引用

IEEE TRANSACTIONS ON neural networks AND LEARNING SYSTEMS 2021年第1期32卷 4-24页

作者： Wu, Zonghan Pan, Shirui Chen, Fengwen Long, Guodong Zhang, Chengqi Yu, Philip S. Univ Technol Sydney Fac Engn & Informat Technol Ctr Artificial Intelligence Ultimo NSW 2007 Australia Monash Univ Fac Informat Technol Clayton Vic 3800 Australia Univ Illinois Dept Comp Sci Chicago IL 60607 USA

Deep learning has revolutionized many machine learning tasks in recent years, ranging from image classification and video processing to speech recognition and natural language understanding. The data in these tasks are typically represented in the Euclidean space. However, there is an increasing number of applications, where data are generated from non-Euclidean domains and are represented as graphs with complex relationships and interdependency between objects. The complexity of graph data has imposed significant challenges on the existing machine learning algorithms. Recently, many studies on extending deep learning approaches for graph data have emerged. In this article, we provide a comprehensive overview of graph neural networks (GNNs) in data mining and machine learning fields. We propose a new taxonomy to divide the state-of-the-art GNNs into four categories, namely, recurrent GNNs, convolutional GNNs, graph autoencoders, and spatial-temporal GNNs. We further discuss the applications of GNNs across various domains and summarize the open-source codes, benchmark data sets, and model evaluation of GNNs. Finally, we propose potential research directions in this rapidly growing field.

关键词： Deep learning neural networks Task analysis Kernel Feature extraction Data mining Learning systems Deep learning graph autoencoder (GAE) graph convolutional networks (GCNs) graph neural networks (GNNs) graph representation learning network embedding

来源：评论

学校读者我要写书评

暂无评论

Hydrogel-Based artificial Synapses for Sustainable Neuromorphic Electronics

引用

ADVANCED MATERIALS 2024年第38期36卷 e2403937页

作者： Yan, Jiongyi Armstrong, James P. K. Scarpa, Fabrizio Perriman, Adam W. Univ Bristol Sch Cellular & Mol Med Bristol BS8 1TD England Univ Bristol Bristol Med Sch Dept Translat Hlth Sci Bristol BS1 3NY England Univ Bristol Bristol Composites Inst Sch Civil Aerosp & Design Engn CADE Univ Walk Bristol BS8 1TR England Australian Natl Univ Res Sch Chem Canberra ACT 2601 Australia Australian Natl Univ John Curtin Sch Med Res Canberra ACT 2601 Australia

Hydrogels find widespread applications in biomedicine because of their outstanding biocompatibility, biodegradability, and tunable material properties. Hydrogels can be chemically functionalized or reinforced to respond to physical or chemical stimulation, which opens up new possibilities in the emerging field of intelligent bioelectronics. Here, the state-of-the-art in functional hydrogel-based transistors and memristors is reviewed as potential artificial synapses. Within these systems, hydrogels can serve as semisolid dielectric electrolytes in transistors and as switching layers in memristors. These synaptic devices with volatile and non-volatile resistive switching show good adaptability to external stimuli for short-term and long-term synaptic memory effects, some of which are integrated into synaptic arrays as artificial neurons;although, there are discrepancies in switching performance and efficacy. By comparing different hydrogels and their respective properties, an outlook is provided on a new range of biocompatible, environment-friendly, and sustainable neuromorphic hardware. How potential energy-efficient information storage and processing can be achieved using artificial neural networks with brain-inspired architecture for neuromorphic computing is described. The development of hydrogel-based artificial synapses can significantly impact the fields of neuromorphic bionics, biometrics, and biosensing. In this review, the current state-of-the-art in functional hydrogel-based transistors and memristors and their potential as artificial synapses are discussed. In doing so, an outlook on new biocompatible and sustainable neuromorphic hardware is provided and how potential energy-efficient information storage and processing can be achieved using artificial neural networks with brain-inspired architecture for neuromorphic computing is discussed. image

关键词： artificial synapses hydrogels memristors neuromorphic computing resistive switching transistors

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：