检索结果-内蒙古大学图书馆

IEEE Transactions on artificial intelligence 2024年第1期5卷 23-37页

作者： Sun, Chenxi Song, Moxian Cai, Derun Zhang, Baofeng Hong, Shenda Li, Hongyan School of Intelligence Science and Technology Beijing100871 China Peking University Health Science Center National Institute of Health Data Science Institute of Medical Technology Beijing100871 China

A recurrent neural network (RNN) has demonstrated its outstanding ability in sequence tasks and has achieved state of the art in many applications, such as industrial and medical. An echo state network (ESN) is a simple type of RNN and has emerged in the last decade as an alternative to gradient descent training-based RNN. The ESN is practical, conceptually simple, and easy to implement with a strong theoretical ground. It can avoid nonconverging and computationally expensive issues in gradient descent RNN methods. Since the ESN was put forward in 2002, abundant existing works have promoted the progress of ESN, and the recently introduced deep ESN opened the way to uniting the merits of deep learning and reservoir computing. Besides, the combinations of ESNs with other machine learning models have also overperformed baselines in some applications. However, the apparent simplicity of ESNs can sometimes be deceptive. Successfully applying ESNs needs some experience. Thus, we reviewed over 300 related papers and provided a systematic overview for the first time. In this article, we categorize the related methods into classical ESN, DeepESN, and combination. Then, we analyze them from the perspective of network designs and specific applications. Finally, we discuss the challenges and opportunities by proposing open problems and future work. © 2020 IEEE.

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

A momental bound algorithm based on precise stepsize adjustment for training deep neural networks 43

A momental bound algorithm based on precise stepsize adjustm...

引用

43rd Chinese Control Conference, CCC 2024

作者： Zhang, Haixia Li, Dequan Qian, Zhen Anhui University of Science and Technology School of Mathematics and Big Data Huainan232000 China Anhui University of Science and Technology School of Artificial Intelligence Huainan232000 China

ISBN: (纸本)9789887581581

Adaptive algorithms are extensively employed in the field of deep learning owing to their rapid convergence properties. Adam is the most common adaptive algorithm among them. However, it has revealed that Adam has a poor generalization ability. AdaBelief is an algorithm based on Adam with exact stepsize adjustment. It introduces a new second-order momentum that corrects Adam's second-order moments. These algorithms still fail to converge during training due to the presence of instability and extreme learning rates. In this paper, we propose a new adaptive and momental bounded algorithm, called AdaBMod, which can effectively mitigate the sudden large learning rate problem and is especially suitable for training deep neural networks. By setting an adaptive finite learning rate in AdaBelief algorithm, the obtained AdaBMod can effectively eliminate the problem of high learning rate in the late training of neural networks, so as to make the training process more stable. Numerical simulation experiments on deep neural network tasks also show that our proposed AdaBMod algorithm eliminates the large learning rate during the training process. The results are also better than other current state-of-the-art optimizers. © 2024 Technical Committee on Control Theory, Chinese Association of Automation.

关键词： Adaptive Algorithms Adaptive and Momental Bound Deep Neural Networks Image Classification Robustness

来源：评论

学校读者我要写书评

暂无评论

Multi-Stage-Based Siamese Neural Network for Seal Image Recognition

引用

computer Modeling in Engineering & sciences 2025年第1期142卷 405-423页

作者： Jianfeng Lu Xiangye Huang Caijin Li Renlin Xin Shanqing Zhang Mahmoud Emam School of Computer Science and Technology Hangzhou Dianzi UniversityHangzhou310018China Shangyu Institute of Science and Engineering Hangzhou Dianzi UniversityShaoxing312300China Faculty of Artificial Intelligence Menoufia UniversityShebin El-Koom32511Egypt

Seal authentication is an important task for verifying the authenticity of stamped seals used in various domains to protect legal documents from tampering and *** seal inspection is commonly audited manually to ensure document ***,manual assessment of seal images is tedious and laborintensive due to human errors,inconsistent placement,and completeness of the *** image recognition systems are inadequate enough to identify seal types accurately,necessitating a neural network-based method for seal image ***,neural network-based classification algorithms,such as Residual Networks(ResNet)andVisualGeometryGroup with 16 layers(VGG16)yield suboptimal recognition rates on stamp ***,the fixed training data categories make handling new categories to be a challenging *** paper proposes amulti-stage seal recognition algorithmbased on Siamese network to overcome these ***,the seal image is pre-processed by applying an image rotation correction module based on Histogram of Oriented Gradients(HOG).Secondly,the similarity between input seal image pairs is measured by utilizing a similarity comparison module based on the Siamese ***,we compare the results with the pre-stored standard seal template images in the database to obtain the seal *** evaluate the performance of the proposed method,we further create a new seal image dataset that contains two subsets with 210,000 valid labeled pairs in *** proposed work has a practical significance in industries where automatic seal authentication is essential as in legal,financial,and governmental sectors,where automatic seal recognition can enhance document security and streamline validation ***,the experimental results show that the proposed multi-stage method for seal image recognition outperforms state-of-the-art methods on the two established datasets.

关键词： Seal recognition seal authentication document tampering siamese network spatial transformer network similarity comparison network

来源：评论

学校读者我要写书评

暂无评论

VaeDiff-DocRE: End-to-end data Augmentation Framework for Document-level Relation Extraction 31

VaeDiff-DocRE: End-to-end Data Augmentation Framework for Do...

引用

31st International Conference on Computational Linguistics, COLING 2025

作者： Tran, Khai Phan Hua, Wen Li, Xue School of Electrical Engineering and Computer Science The University of Queensland Australia Department of Data Science and Artificial Intelligence The Hong Kong Polytechnic University Hong Kong

ISBN: (纸本)9798891761964

Document-level Relation Extraction (DocRE) aims to identify relationships between entity pairs within a document. However, most existing methods assume a uniform label distribution, resulting in suboptimal performance on real-world, imbalanced datasets. To tackle this challenge, we propose a novel data augmentation approach using generative models to enhance data from the embedding space. Our method leverages the Variational Autoencoder (VAE) architecture to capture all relation-wise distributions formed by entity pair representations and augment data for underrepresented relations. To better capture the multi-label nature of DocRE, we parameterize the VAE's latent space with a Diffusion Model. Additionally, we introduce a hierarchical training framework to integrate the proposed VAE-based augmentation module into DocRE systems. Experiments on two benchmark datasets demonstrate that our method outperforms state-of-the-art models, effectively addressing the long-tail distribution problem in DocRE. Our code is released at: https://***/khaitran22/VaeDiff-DocRE. © 2025 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Heterogeneous Device Collaboration Based Federated Learning for Big data Applications

引用

IEEE Transactions on Big data 2024年 1-11页

作者： Wang, Wenhua Yang, Quan Liang, Yuzhu Xu, Yang Liu, Qin Wang, Tian School of Artificial Intelligence Beijing Normal University Beijing China College of Computer Science and Technology Huaqiao University Xiamen China College of Computer Science and Electronic Engineering Hunan University Changsha China BNU-UIC Institute of Artificial Intelligence and Future Networks Beijing Normal University Zhuhai China

In the era of Big data, artificial intelligence and information science are the key technologies to extract the value of data and enhance the competitiveness of enterprises. The characteristics of distributed, small-scale, and sparse lead to the isolated data island problem. To solve these problems, Federated Learning is proposed. However, a large number of terminal models need to be uploaded to the server in Federated Learning, especially for the actual scenario of Internet of Things. Therefore, huge communication costs are required which dramatically increases the pressure on the backbone network. Furthermore, the low quality of the local model will lead to decreased accuracy and convergence rates of the model. To overcome the above limitations, we propose heterogeneous device collaboration based federated learning (HDCFL), which constructs a three-layer structure for Federated Learning by leveraging edge computing and designs a heterogeneous device collaboration method that groups the terminals based on their computing power, communication time, and data volume to train the model. Then, we conduct a theoretical analysis of the proposed algorithm which verifies its advantage. At last, the experimental result demonstrates that the proposed algorithm consistently achieves superior performance in terms of both convergence speed and accuracy compared with state-of-the-art baselines. IEEE

关键词： Competition

来源：评论

学校读者我要写书评

暂无评论

A CT Image Denoising Approach Using Wiener Filter and NSST Domain

A CT Image Denoising Approach Using Wiener Filter and NSST D...

引用

2024 International Conference on artificial intelligence and Emerging technology, Global AI Summit 2024

作者： Katta, Swapna Singh, Prabhishek Garg, Deepak School of Computer Science and Artificial Intelligence Sr University Telangana Warangal India School of Computer Science Engineering and Technology Bennett University Greater Noida India

ISBN: (纸本)9798350379716

Denoising (LDCT) low dose computed-tomographical imaging is essential for improving diagnostic preciseness and image clarity, especially in the presence of gaussian noise. In LDCT imaging, gaussian noise adversely affects the image quality. To address this concern, the proposed a novel method combines the (NSST) no subsampled shearlet transform with wiener filtering and bivariate shrinkage, resulting in improved denoising outcomes. Initially, the NSST approach, decomposes the CT image into approximation and detail parts. In the approximation part, the wiener filter was applied, and bivariate shrinkage thresholding was applied in the detail part to get denoised coefficients. The inverse shearlet transform is employed for obtaining the final-denoised image. The experimental outcome of the proposed method is enumerated relating to the (PSNR) peak signal-to-noise ratio, (SNR) signal to noise ratio, and (UIQI) universal image quality index of particular CT images. The PSNR, SNR, and UIQI values at different noise variances (σ = 10, 15, 20, and 25) have been taken to analyze the performance of quantitative metrics with standard denoising filters. Here, the proposed method provides better results with regards to noise suppression and visual quality of CT images. © 2024 IEEE.

关键词： Wiener filtering

来源：评论

学校读者我要写书评

暂无评论

Multimodal diagnosis of Alzheimer’s disease based on volumetric and cognitive assessments

引用

Multimedia Tools and Applications 2025年第17期84卷 18223-18246页

作者： Kumari, Rashmi Das, Subhranil Nigam, Akriti Singh, Raghwendra Kishore School of Computer Science Engineering and Technology Bennett University Uttar Pradesh Greater Noida201310 India Uttarakhand Dehradun248007 India Birla Institute of Technology Mesra Jharkhand Ranchi835215 India School of Computer Science and Artificial Intelligence SR University Anantasagar Telangana Hasanparthy506371 India

Alzheimer’s Disease (AD) is the most prevalent form of dementia, severely impacting memory functions. Early classification of AD patients, distinguishing them from those with Mild Cognitive Impairment (MCI) or normal cognition (NC), is crucial for clinical intervention. However, traditional methods relying on single MRI feature evaluation yield suboptimal accuracy. Recent advances in multimodal studies show that combining multiple features from various T1-weighted MRI (T1-W1) images can significantly enhance classification accuracy. In this study, we propose a novel Decision Classification Tree (DCT) technique applied to T1-weighted MRI images, incorporating cognitive assessment for global volume measurement. It utilizes morphometric features and spatial information from different Regions of Interest (ROI) within localized brain areas, improving classification accuracy. We evaluate DCT against seven state-of-the- art techniques for binary classification categories (AD vs MCI, MCI vs CN, and AD vs CN) using metrics like accuracy, sensitivity, specificity, and ROC curve analysis. Testing on the ADNI dataset with 102 subjects, we achieve an accuracy of 91.6%, 100% specificity, and 83% sensitivity for the AD vs CN binary class. For MCI vs AD, we attain 76% accuracy, 77.2% specificity, and 80% sensitivity. For MCI vs CN, our model achieves 80% accuracy, 89.4% specificity, and 80% sensitivity. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Magnetic resonance imaging

来源：评论

学校读者我要写书评

暂无评论

A Clearer Image: Improving Object Detection in Real Rainy Conditions with Two-Stage Processing

A Clearer Image: Improving Object Detection in Real Rainy Co...

引用

2023 IEEE International Conference on Multimedia and Expo Workshops, ICMEW 2023

作者： Yang, Mingxuan Han, Xiyu Ping, Xianyao Li, Zipeng Xiao, Jing School of Computer Science and Artificial Intelligence Wuhan University of Technology China School of Computer Science Wuhan University China

ISBN: (纸本)9798350313154

Object detection is one of the most important and challenging branches of computer vision, which has been widely applied in social life, such as intelligent security, autonomous driving, and so on. However, the performance of object detection could degrade rapidly under challenging weather scenarios including rainy conditions. Despite existing technologies having made impressive advancements, they are almost under synthetic rainy-clean pairs, which have a huge domain gap between synthetic and real rain. In this work, we propose an unsupervised deraining method under real rain and utilize a non-local contrastive learning for better decoupling the rain layer from the clean image. Furthermore, we introduce a semi-supervised detector during detection module to reduce the inconsistency of pseudo labels for better facilitating contrastive learning. Extensive experiments with different deraining and object detection methods demonstrate that our two-stage method obtains competitive performance in complex weather scenarios. © 2023 IEEE.

关键词： Rain

来源：评论

学校读者我要写书评

暂无评论

A Domain Generalization and Residual Network-Based Emotion Recognition from Physiological Signals

引用

类生命系统（英文） 2024年第6期2024卷 423-434页

作者： Junnan Li Jiang Li Xiaoping Wang Xin Zhan Zhigang Zeng School of Artificial Intelligence and Automation Huazhong University of Science and TechnologyWuhanChina Hubei Key Laboratory of Brain-inspired Intelligent Systems Huazhong University of Science and TechnologyWuhanChina Key Laboratory of Image Processing and Intelligent Control(Huazhong University of Science and Technology) Ministry of EducationWuhanChina School of Artificial Intelligence and Automation Huazhong University of Science and TechnologyWuhanChina Hubei Key Laboratory of Brain-inspired Intelligent Systems Huazhong University of Science and TechnologyWuhanChina Key Laboratory of Image Processing and Intelligent Control(Huazhong University of Science and Technology) Ministry of EducationWuhanChina Institute of Artificial Intelligence Huazhong University of Science and TechnologyWuhanChina

Emotion recognition from physiological signals(ERPS)has drawn tremendous attention and can be potentially applied to numerous *** physiological signals are nonstationary time series with high sampling frequency,it is challenging to directly extract features from ***,there are 2 major challenges in ERPS:(a)how to adequately capture the correlations between physiological signals at different times and between different types of physiological signals and(b)how to effectively minimize the negative effect caused by temporal covariate shift(TCS).To tackle these problems,we propose a domain generalization and residual network-based approach for emotion recognition from physiological signals(DGR-ERPS).We first pre-extract time-and frequency-domain features from the original time series to compose a new time ***,in order to fully extract the correlation information of different physiological signals,these time series are converted into 3D image data to serve as input for a residual-based feature encoder(RBFE).In addition,we introduce a domain generalization-based technique to mitigate the issue posed by *** have conducted extensive experiments on 2 real-world datasets,and the results indicate that our DGR-ERPS achieves superior performance under both TCS and non-TCS scenarios.

关键词： emotion domain from generalization network-based physiological recognition residual signals

来源：评论

学校读者我要写书评

暂无评论

T5-LDSum: Leveraging T5 Transformer in Hybrid Abstractive-Extractive Long Document Summarization 6

T5-LDSum: Leveraging T5 Transformer in Hybrid Abstractive-Ex...

引用

6th IEEE Novel Intelligent and Leading Emerging sciences Conference, NILES 2024

作者： Magdy, Mostafa Abdelbaky, Abdelrahman M. Mohamed, Ensaf Hussein Helwan University Master of Computer Science Faculty of Computers and Artificial Intelligence Egypt Helwan University Bachelor of Computer Science Faculty of Computers and Artificial Intelligence Egypt Nile University Computer Science School of Information Technology and Computer Science Giza Egypt Helwan University Faculty of Computers and Artificial Intelligence Computer Science Department Egypt

ISBN: (纸本)9798350378511

In the era of digital information overload, the ability to summarize books efficiently emerges as an invaluable skill. Book summarization condenses extensive texts into digestible, concise summaries, enabling readers to grasp the essence of a book without committing to reading it in its entirety. In scholarly research, various techniques for text summarization are employed, including extractive summarization, abstractive summarization, and hybrid approaches. This paper presents a novel hybrid method that leverages the strengths of both extractive and abstractive techniques, with a particular focus on the T5 text-to-text transformer model. Our methodology begins with a BERT-based extractive model paired with the LexRank algorithm to generate preliminary summaries. These summaries are subsequently refined using the T5 transformer model, known for its powerful text generation capabilities. We conducted two empirical studies using Oscar Wilde's 'The Picture of Dorian Gray,' specifically focusing on Chapter 13. Our experiments analyzed the efficacy of summaries derived from the top 20% and top 60% of sentences ranked by the extractive model. The findings indicate a consistent pattern: summaries based on the 60% sentence extraction significantly outperform those from the 20% extraction across all assessed metrics-Rouge, precision, recall, F1 scores, and human evaluations. The integration of the T5 transformer model in the refinement process is highlighted as a key component in achieving high-quality, coherent summaries. © 2024 IEEE.

关键词： abstractive summarization Book summarization evaluation metrics extractive summarization hybrid summarization

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：