检索结果-内蒙古大学图书馆

TransXNet: Learning Both Global and Local Dynamics With a Dual Dynamic Token Mixer for Visual Recognition

IEEE Transactions on Neural Networks and Learning Systems 2025年第6期36卷 11534-11547页

作者： Meng Lou Shu Zhang Hong-Yu Zhou Sibei Yang Chuan Wu Yizhou Yu School of Computing and Data Science The University of Hong Kong Hong Kong SAR China Artificial Intelligence Laboratory Deepwise Healthcare Beijing China Department of Biomedical Informatics Harvard Medical School Boston CA USA Department of Computer Science The University of Hong Kong Hong Kong SAR China School of Information Science and Technology ShanghaiTech University Shanghai China

Recent studies have integrated convolutions into transformers to introduce inductive bias and improve generalization performance. However, the static nature of conventional convolution prevents it from dynamically adapting to input variations, resulting in a representation discrepancy between convolution and self-attention as self-attention calculates attention matrices dynamically. Furthermore, when stacking token mixers that consist of convolution and self-attention to form a deep network, the static nature of convolution hinders the fusion of features previously generated by self-attention into convolution kernels. These two limitations result in a suboptimal representation capacity of the constructed networks. To find a solution, we propose a lightweight dual dynamic token mixer (D-Mixer) to simultaneously learn global and local dynamics, that is, mechanisms that compute weights for aggregating global contexts and local details in an input-dependent manner. D-Mixer works by applying an efficient global attention module and an input-dependent depthwise convolution separately on evenly split feature segments, endowing the network with strong inductive bias and an enlarged effective receptive field. We use D-Mixer as the basic building block to design TransXNet, a novel hybrid CNN–transformer vision backbone network that delivers compelling performance. In the ImageNet-1K image classification task, TransXNet-T surpasses Swin-T by 0.3% in top-1 accuracy while requiring less than half of the computational cost. Furthermore, TransXNet-S and TransXNet-B exhibit excellent model scalability, achieving top-1 accuracy of 83.8% and 84.6%, respectively, with reasonable computational costs. In addition, our proposed network architecture demonstrates strong generalization capabilities in various dense prediction tasks, outperforming other state-of-the-art networks while having lower computational costs. Code is publicly available at https://***/LMMMEng/TransXNet .

关键词： Mixers Transformers Computational modeling Kernel Computational efficiency Feature extraction Convolutional codes Computer vision Accuracy Stacking

来源：评论

学校读者我要写书评

暂无评论

MultiSFL: Towards Accurate Split Federated Learning via Multi-Model Aggregation and Knowledge Replay 39

MultiSFL: Towards Accurate Split Federated Learning via Mult...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Xia, Zeke Hu, Ming Yan, Dengke Liu, Ruixuan Li, Anran Xie, Xiaofei Chen, Mingsong MoE Engineering Research Center of SW/HW Co-Design Tech. and App. East China Normal University China School of Computing and Information Systems Singapore Management University Singapore Department of Biomedical Informatics & Data Science School of Medicine Yale University United States

ISBN: (纸本)157735897X

Although Split Federated Learning (SFL) effectively enables knowledge sharing among resource-constrained clients, it suffers from low training performance due to the neglect of data heterogeneity and catastrophic forgetting problems. To address these issues, we propose a novel SFL approach named MultiSFL, which adopts i) an effective multi-model aggregation mechanism to alleviate gradient divergence caused by heterogeneous data and ii) a novel knowledge replay strategy to deal with the catastrophic forgetting problem. MultiSFL adopts two servers (i.e., the fed server and main server) to maintain multiple branch models for local training and an aggregated master model for knowledge sharing among branch models. To mitigate catastrophic forgetting, the main server of MultiSFL selects multiple assistant devices for knowledge replay according to the training data distribution of each full branch model. Experimental results obtained from various non-IID and IID scenarios demonstrate that MultiSFL significantly outperforms conventional SFL methods by up to a 23.25% test accuracy improvement. © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

Integration of IoMT Sensors’ Data from Mobile Applications into Cloud Based Personal Health Record 13th

Integration of IoMT Sensors’ Data from Mobile Applications ...

引用

13th International Conference on ICT Innovations, ICT Innovations 2021

作者： Savoska, Snezana Blazeska-Tabakovska, Natasha Jolevski, Ilija Bocevska, Andrijana Ristevski, Blagoj Kilintzis, Vassilis Chatzis, Vagelis Beredimas, Nikolaos Maglaveras, Nicos Trajkovik, Vladimir Faculty of Information and Communication Technologies – Bitola University "St. Kliment Ohridski" – Bitola ul. Partizanska bb Bitola7000 Macedonia Laboratory of Computing Medical Informatics and Biomedical Imaging Technologies Aristotle University Thessaloniki Greece Faculty of Computer Science and Engineering "Ss. Cyril and Methodius" University Skopje Skopje Macedonia

ISBN: (纸本)9783031042058

Mobile applications for vital signs measurement are popular in ambient assisted living environments. They typically include various wearables and devices with their applications connected with the production company with some security preferences. When a manufacturer develops and markets a specific sensor, a smartphone application is developed. Many of the sensors and devices are secured and can work only with its host application developed by the manufacturer. However, for these sensors and devices related to a patient’s cloud-based Personal Health Record (PHR), different mobile applications must be created to connect all additional wearables or IoT-based medical sensors or devices to the appropriate cloud-based PHR. All data from these mobile applications needs be collected into the PHR database securely and in time to store medical information for the patient's health in the PHR. This paper presents the integration of two types of mobile applications – for medical professionals and citizens with the PHR, intended for the Cross4all project. The paper explains how the mobile application is integrated with different sensors and devices and the cloud PHR. We also describe the challenges that have arisen from the pilot project implementation and solutions. The presented concept improves cross-border evidence-based healthcare and integrates the e-prescription and e-referral system intended to solve some cross-border healthcare problems for foreign citizens. © 2022, Springer Nature Switzerland AG.

关键词： Mobile computing

来源：评论

学校读者我要写书评

暂无评论

Employing texture features of chest x-ray images and machine learning in covid-19 detection and classification

Mendel

引用

Mendel 2021年第1期27卷 9-17页

作者： Alquran, Hiam Alsleti, Mohammad Alsharif, Roaa Qasmieh, Isam Abu Alqudah, Ali Mohammad Harun, Nor Hazlyna Binti Department of Biomedical Systems and Informatics Engineering Yarmouk University Irbid21163 Jordan The Institute of Biomedical Technology King Hussein Medical Center Royal Jordanian Medical Service Amman11855 Jordan College of applied medical Sciences Radiological Science Program King Saud University Jeddah21435 Saudi Arabia Data Science Research Lab School of Computing Universiti Utara Malaysia Sintok Kedah06010 Malaysia

The novel coronavirus (nCoV-19) was first detected in December 2019. It had spread worldwide and was declared coronavirus disease (COVID-19) pandemic by March 2020. Patients presented with a wide range of symptoms affecting multiple organ systems predominantly the lungs. Severe cases required intensive care unit (ICU) admissions while there were asymptomatic cases as well. Although early detection of the COVID-19 virus by Real-time reverse transcription-polymerase chain reaction (RT-PCR) is effective, it is not efficient;as there can be false negatives, it is time consuming and expensive. To increase the accuracy of in-vivo detection, radiological image-based methods like a simple chest X-ray (CXR) can be utilized. This reduces the false negatives as compared to solely using the RT-PCR technique. This paper employs various image processing techniques besides extracted texture features from the radiological images and feeds them to different artificial intelligence (AI) scenarios to distinguish between normal, pneumonia, and COVID-19 cases. The best scenario is then adopted to build an automated system that can segment the chest region from the acquired image, enhance the segmented region then extract the texture features, and finally, classify it into one of the three classes. The best overall accuracy achieved is 93.1% by exploiting Ensemble classifier. Utilizing radiological data to conform to a machine learning format reduces the detection time and increase the chances of survival. © 2021, Brno University of Technology. 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

Long-Tailed Federated Learning in Internet of Medical Things Based on Ensemble Distillation and Imbalanced Calibration

引用

IEEE Transactions on Consumer Electronics 2025年

作者： Jiang, Bin Shang, Yuchen Yue, Guanghui Wang, Huihui Helen Song, Houbing Herbert College of Oceanography and Space Informatics Qingdao266580 China Shenzhen University School of Biomedical Engineering Health Science Center Shenzhen518060 China Northeastern University Computing Programs 1300 17th St N 1500 ArlingtonVA22209 United States Department of Information Systems BaltimoreMD21250 United States

The Internet of Medical Things (IoMT) has a promising future, as its devices can monitor vital signs, offer treatment guidance, and perform real-time diagnostics using AI and wireless communication technologies. However, due to the difficulty of collecting patient data on a large scale and potential privacy risks, traditional centralized machine learning methods are often challenging to apply in IoMT devices. Federated learning, as a privacy-preserving technology, aims to build high-quality deep learning models across distributed clients while protecting data privacy. However, current popular federated learning methods exhibit suboptimal performance when dealing with non-IIDness data, especially in the case of long-tail class distributions, leading to unsatisfactory results. Additionally, due to privacy constraints on distributed clients, these methods cannot leverage traditional deep learning techniques to handle long-tail data, which is often characterized by long-tail heterogeneous distributions in IoMT. To address these challenges, this paper proposes a solution of Privacy-preserving computing Client Scoring and Knowledge Distillation (FedLT+SKD). The method uses privacy protection computation to provide prior knowledge of global data class distribution while ensuring data privacy. Based on this prior knowledge, it employs a points-based sampling strategy to identify clients that perform well on long tail data and uploads their local model to the server. On the server side, the robustness of the global model is enhanced by collection distillation and imbalance correction. We verify the effectiveness of this method on the medical datasets ISIC, ChestX-ray14, MRI and also on the traditional datasets CIFAR-10-LT and CIFAR-100-LT, and the experimental results show that the method is superior to the popular federation and long-tail learning methods. © 1975-2011 IEEE.

关键词： Differential privacy

来源：评论

学校读者我要写书评

暂无评论

CT-based Anomaly Detection of Liver Tumors Using Generative Diffusion Prior

arXiv

引用

arXiv 2024年

作者： Shi, Yongyi Niu, Chuang Simpson, Amber L. De Man, Bruno Do, Richard Wang, Ge Biomedical Imaging Center Department of Biomedical Engineering School of Engineering Rensselaer Polytechnic Institute TroyNY12180 United States Biomedical Computing and Informatics Queen’s University ONK7L 3N6 Canada GE HealthCare Technology & Innovation Center NiskayunaNY12309 United States Department of Radiology Memorial Sloan Kettering Cancer Center New YorkNY10065 United States

CT is a main modality for imaging liver diseases, valuable in detecting and localizing liver tumors. Traditional anomaly detection methods analyze reconstructed images to identify pathological structures. However, these methods may produce suboptimal results, overlooking subtle differences among various tissue types. To address this challenge, here we employ generative diffusion prior to inpaint the liver as the reference facilitating anomaly detection. Specifically, we use an adaptive threshold to extract a mask of abnormal regions, which are then inpainted using a diffusion prior to calculating an anomaly score based on the discrepancy between the original CT image and the inpainted counterpart. Our methodology has been tested on two liver CT datasets, demonstrating a significant improvement in detection accuracy, with a 7.9% boost in the area under the curve (AUC) compared to the state-of-the-art. This performance gain underscores the potential of our approach to refine the radiological assessment of liver diseases. Copyright © 2024, The Authors. All rights reserved.

关键词： Computerized tomography

来源：评论

学校读者我要写书评

暂无评论

To Improve Supportive Care for Patients Taking Oral Anticancer Agents 18

To Improve Supportive Care for Patients Taking Oral Anticanc...

引用

18th World Congress on Medical and Health informatics: One World, One Health - Global Partnership for Digital Innovation, MEDINFO 2021

作者： He, Lu Song, Ting Jiang, Yun Yu, Ping Song, Lixin Gong, Yang Department of Informatics University of California IrvineCA United States School of Computing and Information Technology University of Wollongong WollongongNSW Australia University of Michigan School of Nursing Ann ArborMI United States School of Nursing & Lineberger Comprehensive Cancer Center University of North Carolina Chapel HillNC United States School of Biomedical Informatics University of Texas Health Science Center 7000 Fannin St HoustonTX77030 United States

ISBN: (纸本)9781643682648

Oral anticancer agents (OAA) are increasingly prescribed to treat cancer because they are flexible and convenient to use. However, managing complex OAA regimens and life-threatening toxicities at home can be challenging for patients and their caregivers. It is urgent to better understand the supportive care needs for OAA and develop novel approaches to facilitating self-management and communicating about OAA. Guided by the chronic care model (CCM), we conducted a grounded theory-based study to analyze OAA-related online discussions and potential mHealth interventions. We found that patients and caregivers commonly used the online community to share personal experiences and concerns, exchange emotional and informational support, identify relevant resources, and obtain benefits of peer coaching. The findings deepen the understanding of the needs for OAA self-management and mHealth interventions, contributing to the development of mHealth models to enhance supportive care and improve communication among peer patients and between patients and providers. © 2022 International Medical informatics Association (IMIA) and IOS Press.

关键词： Medical informatics

来源：评论

学校读者我要写书评

暂无评论

Improving Imbalanced Learning by Pre-finetuning with Data Augmentation 4

Improving Imbalanced Learning by Pre-finetuning with Data Au...

引用

4th International Workshop on Learning with Imbalanced Domains: Theory and Applications, LIDTA 2022

作者： Shi, Yiwen Aslani, Taha Valizadeh Wang, Jing Ren, Ping Zhang, Yi Hu, Meng Zhao, Liang Liang, Hualou College of Computing and Informatics Drexel University Philadelphia United States College of Engineering Drexel University Philadelphia United States Office of Research and Standards Office of Generic Drugs Center for Drug Evaluation and Research Food and Drug Administration Silver Spring United States School of Biomedical Engineering Science & Health Systems Drexel University Philadelphia United States

Imbalanced data is ubiquitous in the real world, where there is an uneven distribution of classes in the datasets. Such class imbalance poses a major challenge for modern deep learning, even with the typical class-balanced approaches such as re-sampling and re-weighting. In this work, we introduced a simple training strategy, namely pre-finetuning, as a new intermediate training stage in between the pretrained model and finetuning. We leveraged the idea of data augmentation to learn an initial representation that better fits the imbalanced distribution of the domain task during the pre-finetuning stage. We tested our method on manually contrived imbalanced datasets (both two-class and multi-class) and the FDA drug labeling dataset for ADME (i.e., absorption, distribution, metabolism, and excretion) classification. We found that, compared with standard single-stage training (i.e., vanilla finetuning), our method consistently attains improved model performance by large margins. Our work demonstrated that pre-finetuning is a simple, yet effective, learning strategy for imbalanced data. © 2022 Y. Shi, T. ValizadehAslani, J. Wang, P. Ren, Y. Zhang, M. Hu, L. Zhao & H. Liang.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

Accurate Estimation of Individual Transmission Rates Through Contact Analytics Using UWB Based Indoor Location Data

Accurate Estimation of Individual Transmission Rates Through...

引用

International Conference on Smart Communications and Networking (SmartNets)

作者： Ravi Chandra Thota Sarawat Murtaza Sara Md Yusuf Sarwar Uddin Majid Bani-Yaghoub Gary Sutkin Department of Computer Science University of Missouri-Kansas City Kansas City USA Division of Computing Analytics and Mathematics University of Missouri-Kansas City Kansas City USA Department of Biomedical and Health Informatics University of Missouri-Kansas City University Health Kansas City Kansas City USA

ISBN: (数字)9798350385328

ISBN: (纸本)9798350385335

Effective control and prevention of infectious dis-eases often require detailed social network data categorized by individual movements and contact structures. Social network data is primarily accessible on a macro scale through sources such as RFID, Bluetooth, and mobile location data. This has limited our ability to accurately estimate transmission rates and identify significant patterns of disease spread within social gatherings. This study aims to address this gap by formulating a methodology that enables precise estimation of the disease transmission rates of each individual. Our methodology consists of the following steps (i) Installing a real-time location system and collection of indoor location data from two IRB-approved social events for the college students; (ii) Estimating infection probabilities as a function of distance between individuals calibrated with COVID-19 transmission data; (iii) Conducting a comprehensive analysis of contacts, including contact events, intensity, and duration; (iv) Utilizing the insights from the preceding steps to compute infection transmission rates for each participant. Our study revealed that the majority of contacts among participants are transient, typically lasting between 1 to 30 seconds within a proximity of 6 feet. However, infection transmission is associated with different types of contact durations and frequency, all hap-pening within a 6-foot area. Furthermore, a comparative analysis of the data from two indoor social events indicates a strong correlation between the duration of contacts and the nature of the social gathering (e.g., student orientation session versus informal meeting), alongside the density of participants (e.g., crowded versus sparsely populated environments). The proposed methodology is the first step toward accurately estimating in-dividual transmission rates, leveraging individual characteristics and the nature of the social event as key determinants. This will potentially enable epidemiolog

关键词： Analytical models Accuracy Social networking (online) Infectious diseases Sociology Estimation Real-time systems

来源：评论

学校读者我要写书评

暂无评论

Classification of Lung Cancer by Using Machine Learning Algorithms 5

Classification of Lung Cancer by Using Machine Learning Algo...

引用

5th International Conference on Engineering Technology and its Applications, IICETA 2022

作者： Al-Tawalbeh, Joman Alshargawi, Baraa Alquran, Hiam Al-Azzawi, Waleed Mustafa, Wan Azani Alkhayyat, Ahmed Jordan University of Science and Technology Department of Biomedical Engineering Irbid21163 Jordan Yarmouk University Biomedical Systems and Medical Informatics Engineering Irbid21163 Jordan Al-Farahidi University Medical Technical College Baghdad Iraq Faculty of Electrical Engineering Technology Pauh Putra Campus Perlis Arau02600 Malaysia Pauh Putra Campus Advanced Computing Centre of Excellence Pauh Putra Campus Perlis Arau02600 Malaysia The Islamic University Faculty of Engineering Najaf Iraq

ISBN: (纸本)9781665472159

Due to the structure of cancer cells, where the majority of the cells are overlapped with each other, early diagnosis of lung cancer is a difficult challenge, so the cause of lung cancer remains unknown and prevention is difficult, early discovery of lung cancer is the only approach to cure it. The classification of lung cancer is a crucial process, based on signs that appear on patients;cancers could easily be predicted and treated. This paper uses KNN, SVM, Naïve Bayes and narrow neural network (NNN) classifiers. KNN showed 85.87% accuracy. SVM, Naïve Bayes and NNN showed 92.6%, 90.3% and 90% accuracy respectively. Results came after the fact that among the four classifiers SVM was the most accurate and much better to classify our data. © 2022 IEEE.

关键词： Diseases

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：