检索结果-内蒙古大学图书馆

2022 IEEE International Conference on Multimedia and Expo, ICME 2022

作者： Che, Jieyong Wang, Chong Chen, Weiwei Dai, Xinmiao Wang, Jun Wu, Jiafei Ningbo University Faculty of Electrical Engineering and Computer Science China School of Information and Control Engineering China University of Mining and Technology China SenseTime Research China

ISBN: (数字)9781665485630

ISBN: (纸本)9781665485630

Deep learning networks are gradually deployed in edge applications, such as phones and cameras, which has a restriction of the computational resources. Thus, to improve the computational efficiency, numerous types of group convolution-based frameworks have been studied, including the dynamic group convolution (DGC). However, it is worth noting that sometimes the same channels are selected in different dynamic heads in DGC, which violates its original intention. It indicates the number of dynamic heads exceeds what is really needed. In this paper, a novel model by pruning dynamic group convolution with static substitute is proposed. Specifically, those channels pruned in the dynamic head can be reselected in the static head of the group convolution in the proposed model. In addition, a new polarization regularization is introduced to prune more useless channels with less accuracy loss. The experiment results on two image classification benchmarks (CIFAR-I00 and TinylmageNet-200) show promising performance. © 2022 IEEE.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

A Novel Transfer Learning Framework for Diagnosing Covid-19 from Chest X-Ray Images with Minimal Computational Resources

SSRN

引用

SSRN 2023年

作者： Abdelwahed, Ahmed Helmi, Ahmed M. Issa, Mohamed Computer and Systems Department Faculty of Engineering Zagazig University Egypt Computer Engineering Dept. Engineering and Information Technology College Buraydah Private Colleges Saudi Arabia Faculty of Computer Science and Information Technology Egypt−Japan University for Science and Technology Egypt

Deep learning proved effective in detecting COVID-19 from chest X-ray images. Although computing is the current bottleneck in training and deploying models, few studies focus on developing efficient frameworks while many aim only at performance. We propose a novel framework for detecting COVID-19 cases from chest X-ray images efficiently. Our study utilizes four high-performing architectures, including MobileNetV3, RegNetX, NasNetMobile, and EfficientNetV2, which use minimal resources, to extract features from X-ray images and classify COVID-19 cases using various statistical models and machine learning algorithms. We evaluated our proposed framework using metrics such as accuracy, precision, recall, specificity, f1-score, and AUC scores. The false negative rate for each feature extractor and classifier pair is reported, and the average inference time, to measure resource efficiency, is calculated. The results show that EfficientNetV2 was the best-performing feature extractor, while MobileNetV3 had the best average inference time and the lowest false negative rate. The Neural Network classifier performed the best, with state-of-the-art accuracy reaching 99.28%, followed by both RBF-SVM and XGBoost. The proposed framework is more robust, reliable, and resource-efficient for COVID-19 detection compared to existing literature. Our findings indicate the prospects of using the proposed framework in efficient clinical diagnosis. © 2023, The Authors. All rights reserved.

关键词： COVID-19

来源：评论

学校读者我要写书评

暂无评论

Semi-Supervised Feature Selection With Multi-Scale Fuzzy Information Fusion: From Both Global and Local Perspectives

引用

IEEE Transactions on Fuzzy Systems 2025年第6期33卷 1825-1839页

作者： Zhou, Nan Liao, Shujiao Chen, Hongmei Ding, Weiping Lu, Yaqian Minnan Normal University School of Mathematics and Statistics Fujian Key Laboratory of Granular Computing and Applications Zhangzhou363000 China Southwest Jiaotong University School of Computing and Artificial Intelligence National Engineering Laboratory of Integrated Transportation Big Data Application Technology Chengdu611756 China Nantong University School of Artificial Intelligence and Computer Science Nantong226019 China City University of Macau Faculty of Data Science 999078 China

In reality, the laborious nature of label annotation leads to the widespread existence of limited labeled data. Moreover, multi-scale data have received widespread attention due to its rich knowledge representation. However, current research on multi-scale data primarily focuses on supervised learning environments, while semi-supervised feature selection for multi-scale data with limited labels remains inadequately explored. Meanwhile, existing studies on multi-scale data often emphasize selecting optimal scales and further conducting feature selection, but this strategy may result in losing potentially valuable information from other scales. To overcome these limitations, by adopting a multi-scale fuzzy information fusion mechanism, this article proposes two new semi-supervised feature selection approaches for multi-scale data with limited labels from both global and local perspectives. Initially, by fusing fuzzy information at various scales, a label learning method is proposed to convert the decision of missing labels into a fuzzy decision of label distribution. Subsequently, label-distributed multi-scale fuzzy global and local rough sets are constructed from global and local perspectives, respectively. Based on the two models, two semi-supervised feature selection algorithms are developed based on global and local multi-scale fuzzy information fusion. Experimental results demonstrate that, compared to other advanced feature selection algorithms, the two proposed algorithms can effectively handle multi-scale data with limited labels and exhibit superior classification performance, computational efficiency and robustness. © 1993-2012 IEEE.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Adaptive Online Non-stochastic Control

arXiv

引用

arXiv 2023年

作者： Mhaisen, Naram Iosifidis, George Faculty of Electrical Engineering Mathematics and Computer Science Delft University of Technology Netherlands

We tackle the problem of Non-stochastic Control (NSC) with the aim of obtaining algorithms whose policy regret is proportional to the difficulty of the controlled environment. Namely, we tailor the Follow The Regularized Leader (FTRL) framework to dynamical systems by using regularizers that are proportional to the actual witnessed costs. The main challenge arises from using the proposed adaptive regularizers in the presence of a state, or equivalently, a memory, which couples the effect of the online decisions and requires new tools for bounding the regret. Via new analysis techniques for NSC and FTRL integration, we obtain novel disturbance action controllers (DAC) with sublinear data adaptive policy regret bounds that shrink when the trajectory of costs has small gradients, while staying sub-linear even in the worst case. © 2023, CC BY.

关键词： Stochastic systems

来源：评论

学校读者我要写书评

暂无评论

Toward Enhancing Cross-Lingual Domain Knowledge Sharing and Transferring for Multilingual Domain Adaptation in NMT

IEEE Transactions on Audio, Speech and Language Processing

引用

IEEE Transactions on Audio, Speech and Language Processing 2025年 33卷 2125-2138页

作者： Zhengtao Yu Jinlei Xu Xiang Huang Zhongfen Deng Yonghua Wen Li Sun Yuxin Huang Hao Peng Philip S. Yu Faculty of Information Engineering and Automation Kunming University of Science and Technology Kunming China Yunnan Province Key Laboratory of Artificial Intelligence Kunming University of Science and Technology Kunming China School of Cyber Science and Technology Beihang University Beijing China Department of Computer Science University of Illinois Chicago IL USA School of Control and Computer Engineering North China Electric Power University Beijing China

Multilingual Neural Machine Translation (NMT) excels in sharing knowledge across languages and transferring insights from high-resource languages to improve performance in low-resource languages. However, its performance lags in specific domains such as legal and medical. Previous works have focused on adding language-specific and domain-specific adapters to achieve domain adaptation. Although effective, these adapter-based methods only use domain data to train additional parameters, limiting the performance of multilingual NMT. In this paper, we propose CDSTX, a novel approach that achieves robust multilingual domain adaptation without relying on adapters, solely leveraging multilingual models. Specifically, we utilize the pretrained model XLM-R and frozen embeddings to preserve its robust multilingual capabilities and design a two-stage training strategy for domain adaptation. This includes separate training for the decoder and fine-tuning of the entire model, ensuring that the model effectively acquires domain knowledge and correctly represents domain-specific text. Moreover, to better utilize the domain information conveyed implicitly by the training data, we devise special domain tokens at the beginning of the source and target sentences called Source&Target Domain Tags. In addition, back translation is employed to enhance the cross-lingual transfer ability of our approach. Our proposed method is evaluated on two datasets across three translation tasks: Single-domain Multilingual NMT, Multi-domain bilingual NMT, and Multilingual Multi-domain NMT. Notably, in the single-domain multilingual NMT task, CDSTX significantly enhances zero-shot domain translation performance, achieving improvements of up to +30 BLEU points through the utilization of back-translation techniques. Even in the bilingual multi-domain NMT task where specific domain data for the target translation direction is unavailable, our method consistently outperforms all SOTA methods. Moreover, in the

关键词： Training Adaptation models Translation Limiting Law Large language models Training data Neural machine translation Multilingual Speech processing

来源：评论

学校读者我要写书评

暂无评论

A Lottery-based Spherical Evolution Algorithm with Elite Retention Strategy 14

A Lottery-based Spherical Evolution Algorithm with Elite Ret...

引用

14th International Conference on Intelligent Human-Machine Systems and Cybernetics, IHMSC 2022

作者： Li, Jiayi Zhang, Zihang Lei, Zhenyu Yi, Junyan Gao, Shangce University of Toyama Faculty of Engineering Toyama930-8555 Japan Beijing University of Civil Engineering and Architecture Department of Computer Science and Technology Beijing100044 China

ISBN: (数字)9781665461696

ISBN: (纸本)9781665461696

Spherical evolution (SE) is a recently proposed meta-heuristic algorithm. Its special search approach has been proved to be very effective in exploring the search space. SE is very powerful for optimization, but still has room for improvement due to some promising solutions usually fail to survive into the next generation. To alleviate this issue, we innovatively design a novel lottery-based elite retention strategy and propose a lottery elite spherical evolution algorithm (LESE). To verify the effectiveness of the proposed LESE, we experimentally compare it with the original SE algorithm and other representative meta-heuristics algorithms. We use 30 benchmark functions from the IEEE CEC2017 benchmark as the test set for our experiments. The effectiveness of LESE is demonstrated by analyzing the experimental results from perspectives of solution accuracy, convergence speed, solution distribution, and search dynamics. © 2022 IEEE.

关键词： Spheres

来源：评论

学校读者我要写书评

暂无评论

design of Lightweight Cryptography based Deep Learning Model for Side Channel Attacks 33

Design of Lightweight Cryptography based Deep Learning Model...

引用

33rd International Telecommunication Networks and Applications Conference, ITNAC 2023

作者： Ahmed, Amjed Abbas Hasan, Mohammad Kamrul Nafi, Nazmus Shaker Aman, Azana Hafizah Islam, Shayla Fadhil, Saif Aamer Center for Cyber Security Faculty of Information Science and Technology Bangi43600 Malaysia Department of Computer Techniques Engineering Baghdad10011 Iraq Integrated Work Package KBR SouthbankVIC Australia Institute of Computer Science and Digital Innovation UCSI University Kuala Lumpur56000 Malaysia

ISBN: (纸本)9798350317138

Depending on a device's encryption mechanism, a wide variety of tangible details could be exposed. These leaks are used in side-channel analysis, which is used to get keys. Due to deep learning's sensitivity to the characteristics of the data being processed, using such algorithms can significantly improve the accuracy and efficiency of side channel analysis. However, classic neural networks are now used for the vast majority of the work that is being done. When the number of nodes in a network grows, so does the efficiency with which key recovery can function. However, the method's computing complexity grows in direct proportion. Overfitting, inadequate capacity for feature extraction, and inefficient training are all potential issues. In this study, we develop a compact convolutional neural network by enhancing a previously existing combination of neural networks. Novel neural network along with previous neural network both have their own implementations of the side-channel analysis used in comparative trials. Statistically, the new network has better accuracy, quicker convergence, and more robustness. As part of the research, heatmaps were provided as a means of data visualisation. The critical interval concentration is higher and the heat value is higher in the new network. Conventional neural networks, which serve as the foundation for various kinds of neural networks, perform much worse than side channel studies based on feature fusion networks. © 2023 IEEE.

关键词： Efficiency

来源：评论

学校读者我要写书评

暂无评论

Fractional derivative in continuous-time Markov processes and applications to epidemics in networks

引用

Physical Review Research 2025年第1期7卷 013017-013017页

作者： Matteo D'Alessandro Piet Van Mieghem Faculty of Electrical Engineering Mathematics and Computer Science Delft University of Technology P.O. Box 5031 2600 GA Delft The Netherlands

Continuous-time Markov processes are governed by the Chapman-Kolmogorov differential equation. We show that replacing the standard time derivative of the governing equation with a Caputo fractional derivative of order 0<α<1, leads to a fractional differential equation whose solution can describe the state probabilities of a class of non-Markovian stochastic processes. We show that the same state probabilities also solve a system of equations that describe semi-Markov processes in which the sojourn times follow a Mittag-Leffler distribution, contrasting the usual Markov processes with exponentially distributed sojourn times. We apply the fractional framework to the ɛ−SIS epidemic process on any contact graph and we propose a microscopic epidemic description in which infection and curing events follow a Mittag-Leffler distribution and are not independent. We analytically prove that the description exactly solves the fractional extension of the Chapman-Kolmogorov differential equation, and we provide an extensive study of how the dependence between events strongly affects the dynamics of the spreading process. We conclude verifying the proposed framework with Monte Carlo simulations.

关键词： Markov processes

来源：评论

学校读者我要写书评

暂无评论

Enhancing Event Extraction and Impact Analysis with Knowledge Graph Augmented Transformer

Enhancing Event Extraction and Impact Analysis with Knowledg...

引用

International SoC design Conference, ISOCC

作者： Kallaya Songklang Kanabadee Srisomboon Wilaiporn Lee Akara Prayote Department of Computer and Information Science Faculty of Applied Science King Mongkut’s University of Technology North Bangkok Bangkok Thailand Department of Electrical and Computer Engineering Faculty of Engineering King Mongkut’s University of Technology North Bangkok Bangkok Thailand

ISBN: (数字)9798350377088

ISBN: (纸本)9798350377095

Understanding the impact of events reported in news articles plays pivotal roles in various tasks. In this study, we propose a framework for extracting event components from the news and analyzing their multifaceted impact. The framework employs transformers techniques for event extraction and leverages a biography knowledge base to assess the potential impact of events across social, political, economic, and culture (SPEC) dimensions. Our framework achieves an F1-score of 78.12% in event identification as well as significant accuracy score in evaluating the impact levels prediction. By integrating results from event extraction and impact analysis, the level of multifaceted influence of the event along with the event information are effectively visualized. This comprehensive approach enhances understanding of multi-perspective interpretation and supports insights for various applications.

关键词： Economics Accuracy Event detection Biographies Knowledge based systems Data visualization Knowledge graphs Tagging Transformers Data mining

来源：评论

学校读者我要写书评

暂无评论

Deep Learning in Medical Image Analysis for Personalized Medicine

Deep Learning in Medical Image Analysis for Personalized Med...

引用

International Symposium on Electronics in Marine (ELMAR)

作者： Irena Galić Marija Habijan Faculty of Electrical Engineering Computer Science and Information Technology Osijek Osijek Croatia

Deep learning (DL) has transformed the field of medical image processing, enabling unprecedented accuracy and efficiency in various applications. It has been widely utilized for medical image analysis of different anatomical regions. In this article, we provide an overview of commonly used deep learning methods, including convolutional neural networks (CNNs), recurrent neural networks (RNNs), and generative adversarial networks (GANs). We briefly overview their applications in medical image analysis, such as image classification, object detection/localization, segmentation, generation, and registration. We also highlight the strengths and limitations of each method and identify the challenges that still need to be addressed, including the limited availability of annotated data, variability in medical images, and the interpretability issue. Finally, we discuss future research directions, including developing explainable deep learning methods and integrating multi-modal data.

关键词：

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：