检索结果-内蒙古大学图书馆

Tile-size aware bitrate allocation for adaptive 360∘ video streaming

Multimedia Tools and Applications 2025年第13期84卷 12615-12632页

作者： Huang, Jiawei Liu, Mingyue Liu, Jingling Gao, Feng Li, Weihe Wang, Jianxin School of Computer Science and Engineering Central South University Hunan Changsha410083 China School of Informatics The University of Edinburgh Edinburgh United Kingdom

360∘ videos have become increasingly popular recently, but consume much more bandwidth than non-360∘ videos. Usually, 360∘ video streaming partitions the video surface into multiple tiles and encodes the tiles independently to effectively and flexibly use limited link bandwidth. However, current bitrate adaptive algorithms generally aim to maximize the bitrate, rather than perceptual quality, resulting in degradation of user experience. More importantly, we reveal that the distribution of tile size is very skewed, that is, a small number of large tiles consumes more bandwidth than a large number of small tiles, further hurting the overall viewing quality. Therefore, in this paper, we propose a tile-size aware bitrate allocation scheme TSA for adaptive 360∘ video streaming to improve the viewing experience of users. Specifically, TSA cautiously decreases the quality of a few large tiles to allocate more bandwidth to large number of small tiles, thus improving the perceptual quality of overall video, without sacrificing large tiles excessively. Experiments over real-world datasets show that TSA effectively improves V-VMAF by up to 39% compared with several state-of-the-art adaptive bitrate algorithms. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.

关键词： Video streaming

来源：评论

学校读者我要写书评

暂无评论

Vision Transformer for Audio-Based Depression Detection on Multi-Lingual Audio Data 24

Vision Transformer for Audio-Based Depression Detection on M...

引用

7th International Conference on Digital Medicine and Image Processing, DMIP 2024

作者： Pratiwi, Monica Sanjaya, Samuel Ady Department of Computer Engineering Faculty of Engineering and Informatics Universitas Multimedia Nusantara Tangerang Indonesia Department of Information System Faculty of Engineering and Informatics Universitas Multimedia Nusantara Tangerang Indonesia

ISBN: (纸本)9798400709586

Depression has the potential to impact death rates, particularly when it comes to death by suicide. Inadequate diagnosis may result in a delay or unsuitable therapy, which can worsen symptoms of depression. Unaddressed or insufficiently addressed depression can result in deteriorating mental well-being, which includes a higher risk of experiencing suicidal ideation and engaging in self-destructive actions. Voice analysis can be employed to distinguish between individuals with depression and those without depression. However, the research that has been conducted on voice recognition to distinguish between depressed and non-depressed individuals keeps focusing on the use of a single dataset source as input for the classification model that is being developed. This work utilizes the Vision Transformer model and various pre-trained transformers, such as DeiT (Data-Efficient Image Transformer) and Swin Transformer, to detect depression based on voice. The objective was to create a more generic model not restricted to a specific language. Integrating Mel-spectrogram characteristics with a vision transformer-based model can enhance the efficacy of voice recognition models when dealing with multi-language data. The result shows 21% higher accuracy than the previous study that also implemented a cross-dataset test. Copyright © 2024 held by the owner/author(s). Publication rights licensed to ACM.

关键词： Spectrographs

来源：评论

学校读者我要写书评

暂无评论

Security Strategy of Digital Medical Contents Based on Blockchain in Generative AI Model

引用

computers, Materials & Continua 2025年第1期82卷 259-278页

作者： Hoon Ko Marek R.Ogiela Division of Computer Science and Engineering Sunmoon UniversityAsan31460Republic of Korea AGH University of Krakow Cryptography and Cognitive Informatics LaboratoryKrakow30-059Poland

This study presents an innovative approach to enhancing the security of visual medical data in the generative AI environment through the integration of blockchain *** combining the strengths of blockchain and generative AI,the research team aimed to address the timely challenge of safeguarding visual medical *** participating researchers conducted a comprehensive analysis,examining the vulnerabilities of medical AI services,personal information protection issues,and overall security *** multi faceted exploration led to an indepth evaluation of the model’s performance and ***,the correlation between accuracy,detection rate,and error rate was *** analysis revealed insights into the model’s strengths and limitations,while the consideration of standard deviation shed light on the model’s stability and performance *** study proposed practical improvements,emphasizing the reduction of false negatives to enhance detection rate and leveraging blockchain technology to ensure visual data integrity in medical *** blockchain to generative AI-created medical content addresses key personal information protection *** utilizing the distributed ledger system of blockchain,the research team aimed to protect the privacy and integrity of medical data especially medical *** approach not only enhances security but also enables transparent and tamperproof ***,the use of generative AI models ensures the creation of novel medical content without compromising personal information,further safeguarding patient *** conclusion,this study showcases the potential of blockchain-based solutions in the medical field,particularly in securing sensitive medical data and protecting patient *** proposed approach,combining blockchain and generative AI,offers a promising direction toward more robust and secure medical content *** research and advancements

关键词： Digitalmedical content medical diagnostic visualization security analysis generativeAI blockchain vulnerability pattern recognition

来源：评论

学校读者我要写书评

暂无评论

Bias analysis of AI models for undergraduate student admissions

引用

Neural Computing and Applications 2025年第12期37卷 7785-7795页

作者： Van Busum, Kelly Fang, Shiaofen Department of Computer Science and Software Engineering Butler University IndianapolisIN United States Computing and Engineering Indiana University Luddy School of Informatics Indiana University Indianapolis IndianapolisIN United States

Bias detection and mitigation is an active area of research in machine learning. This work extends previous research done by the authors Van Busum and Fang (Proceedings of the 38th ACM/SIGAPP Symposium on Applied Computing, 2023) to provide a rigorous and more complete analysis of the bias found in AI predictive models. Admissions data spanning six years was used to create an AI model to determine whether a given student would be directly admitted into the School of Science under various scenarios at a large urban research university. During this time, submission of standardized test scores as part of a student’s application became optional which led to interesting questions about the impact of standardized test scores on admission decisions. We developed and analyzed AI models to understand which variables are important in admissions decisions, and how the decision to exclude test scores affects the demographics of the students who are admitted. We then evaluated the predictive models to detect and analyze biases these models may carry with respect to three variables chosen to represent sensitive populations: gender, race, and whether a student was the first in his/her family to attend college. We also extended our analysis to show that the biases detected were persistent. Finally, we included several fairness metrics in our analysis and discussed the uses and limitations of these metrics. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Students

来源：评论

学校读者我要写书评

暂无评论

A Large-scale Multimodal Study for Predicting Mortality Risk Using Minimal and Low Parameter Models and Separable Risk Assessment

引用

IEEE Journal of Biomedical and Health informatics 2025年第5期29卷 3762-3771页

作者： Ulloa Cerna, Alvaro Emilio Vanmaanen, David P. Jing, Linyuan Stough, Joshua V. Patel, Aalpen Haggerty, Christopher M. Fornwalt, Brandon K. Pattichis, Marios Department of Translational Data Science and Informatics Geisinger PA17822 United States University of New Mexico Department of Electrical and Computer Engineering AlbuquerqueNM87131 United States

The majority of biomedical studies use limited datasets that may not generalize over large heterogeneous datasets that have been collected over several decades. The current paper develops and validates several multimodal models that can predict 1-year mortality based on a massive clinical dataset. Our focus on predicting 1-year mortality can provide a sense of urgency to the patients. Using the largest dataset of its kind, the paper considers the development and validation of multimodal models based on 25,137,015 videos associated with 699,822 echocardiography studies from 316,125 patients, and 2,922,990 8-lead electrocardiogram (ECG) traces from 631,353 patients. Our models allow us to assess the contribution of individual factors and modalities to the overall risk. Our approach allows us to develop extremely low-parameter models that use optimized feature selection based on feature importance. Based on available clinical information, we construct a family of models that are made available in the DISIML package. Overall, performance ranges from an AUC of 0.72 with just ten parameters to an AUC of 0.89 with under 105 k for the full multimodal model. The proposed approach represents a modular neural network framework that can provide insights into global risk trends and guide therapies for reducing mortality risk. © 2013 IEEE.

关键词： Electronic health record

来源：评论

学校读者我要写书评

暂无评论

Joint V2V Group Association and Transmission Power Control in Federated Vehicular Networks

引用

IEEE Internet of Things Journal 2025年第11期12卷 18167-18182页

作者： Liu, Su Guan, Peiyuan Li, Yushuai Li, Tianyi Zolfagharian, Zolaikha Yu, Jiong Huang, Tingwen Xinjiang University School of Computer Science and Technology Urumqi830046 China University of Oslo Department of Informatics Oslo0316 Norway Aalborg University Department of Computer Science Aalborg9220 Denmark University of Tehran School of Electrical and Computer Engineering Iran Shenzhen University of Advanced Technology Faculty of Computer Science and Control Engineering Shenzhen518055 China

With growing awareness of privacy protection, Federated Learning (FL) in vehicular network scenarios effectively addresses privacy concerns, leading to the development of Federated Vehicular Networks (FVN). In FVN, vehicles maintain a global model by transmitting local models and iterative processing, resulting in significant communication overhead. In the Internet of Vehicles (IoV), as the majority of the spectrum resources are allocated to Vehicle-to-Vehicle (V2V) communication, vehicles engaged in FL encounter uplink interference when these resources are reused. This compromises the efficacy of FL vehicle communication. To address this issue, we propose FedCDC, a new Federated learning framework with adaptive weight Clustering with knowledge Distillation and Channel sharing-based resource allocation. First, we implement a communication compression strategy based on clustering and distillation to alleviate transmission load. Then, we develop a resource allocation strategy to maximize the Signal-to-Interference-plus-Noise Ratio (SINR) for both FL vehicles and V2V groups, which investigates two critical components: channel pairing and power coordination. We formulate the optimization issue as a multi-objective optimization problem that is solved offline using the MIDACO solver. Due to the dynamic characteristics of FVN, we employ a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) to enhance efficiency. Finally, extensive experiments demonstrate that our approach significantly reduces data transmission and offers an efficient resource coordination plan, thus improving FL communication efficiency and ensuring the Quality of Service (QoS) for both V2V groups and FL vehicles. © 2014 IEEE.

关键词： Differential privacy

来源：评论

学校读者我要写书评

暂无评论

The social force model: a behavioral modeling approach for information propagation during significant events

引用

International Journal of computers and Applications 2025年第3期47卷 279-292页

作者： Saleema, A. Thampi, Sabu M. Thiruvananthapuram India National Informatics Centre Kerala State Centre Thiruvananthapuram India School of Computer Science and Engineering Digital University Kerala Kerala Thiruvananthapuram India

Information diffusion otherwise known as the propagation, spread or dissemination of information occurs when a piece of information flows from a particular individual/community to another in a social network. Studies related to information propagation involve problems regarding the factors that affect the information propagation, how the information is disseminated, the speed of propagation, etc. Researchers have proposed information propagation models to understand the phenomenon and to answer these questions. These models have been effectively used in applications such as behavior analysis, public health care, etc. Although several studies are carried out in this field, the literature demands identifying the most influential factors of propagation in real time in cases of sudden unexpected significant disasters/epidemics/pandemics, since the existing propagation models seem unfit during such circumstances. In this paper, a novel information propagation model which predicts the top propagators of information related to a particular context is proposed. This model utilizes the past few weeks' data during a sudden outbreak of a disaster and identifies the most influential attributes of a user profile to predict the top propagators of the future. The proposed Social Force Model is inspired by a model used in studying the fear propagation pattern in pedestrian dynamics in real-life situation [Cornes FE, Frank GA, Dorso CO. Fear propagation and the evacuation dynamics. Simul Model Pract Theory. 2019;95:112–133.]. We have effectively mapped the various forces which constitute the Social Force Model such as the Desired Force, the Social Force and the Granular Force with respect to the online social network context in order to discover the key spreaders of information during a specific context. Apart from identifying the propagators, the proposed model discovers the key attributes by analyzing the behavior of users based on their past activities in the online social networ

关键词： Social networking (online)

来源：评论

学校读者我要写书评

暂无评论

Comparison of Classical and Deep Learning-Based Feature Representations for Age-Related Macular Degeneration 5th

Comparison of Classical and Deep Learning-Based Feature Re...

引用

5th International Conference on Medical Imaging and computer Aided Diagnosis, MICAD 2024

作者： Sinichi, Parsa Bernabeu, Miguel O. Javidi, Malihe Department of Computer Engineering Faculty of Electronic and Computer Engineering Quchan University of Technology Quchan Iran Centre for Medical Informatics University of Edinburgh Edinburgh United Kingdom

ISBN: (纸本)9789819638628

Age-related Macular Degeneration (AMD) is a leading cause of visual impairment among the elderly worldwide. This study compares deep learning-based and classical feature extraction methods for AMD classification using colour fundus images, a cost-effective imaging modality. This comparative approach aims to evaluate the performance of advanced deep learning methods against traditional computer vision approaches in generating effective feature representation. To achieve this, we compared various classical feature extraction methods alongside deep learning-based approaches, specifically VGG16 pre-trained on ImageNet, RETFound and RETFound-Green, which are foundation models pre-trained on a significant number of retinal images. The results demonstrate that combinations of classical features still obtained notable performance and even outperformed the VGG16 model pre-trained on ImageNet. These results also highlight the need for deep learning models to be trained on domain-specific data to surpass classical approaches. In this regard, the pre-trained RETFound and RETFound-Green models yielded promising outcomes for AMD classification, and the highest performance was achieved using a fine-tuned RETFound model. Furthermore, the best proposed method was compared with the state-of-the-art methods. This approach achieved competitive results on the ADAM dataset, demonstrating its robustness and effectiveness in AMD classification while utilising a significantly simpler architecture. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

xCViT:Improved Vision Transformer Network with Fusion of CNN and Xception for Skin Disease Recognition with Explainable AI

引用

computers, Materials & Continua 2025年第4期83卷 1367-1398页

作者： Armughan Ali Hooria Shahbaz Robertas Damaševicius Department of Electrical Engineering Wah Engineering CollegeUniversity of WahWah Cantt47040Pakistan Department of Computer Science HITEC UniversityTaxila47080Pakistan Department of Applied Informatics Vytautas Magnus UniversityKaunas44309Lithuania

Skin cancer is the most prevalent cancer globally,primarily due to extensive exposure to Ultraviolet(UV)*** identification of skin cancer enhances the likelihood of effective treatment,as delays may lead to severe tumor *** study proposes a novel hybrid deep learning strategy to address the complex issue of skin cancer diagnosis,with an architecture that integrates a Vision Transformer,a bespoke convolutional neural network(CNN),and an Xception *** were evaluated using two benchmark datasets,HAM10000 and Skin Cancer *** the HAM10000,the model achieves a precision of 95.46%,an accuracy of 96.74%,a recall of 96.27%,specificity of 96.00%and an F1-Score of 95.86%.It obtains an accuracy of 93.19%,a precision of 93.25%,a recall of 92.80%,a specificity of 92.89%and an F1-Score of 93.19%on the Skin Cancer ISIC *** findings demonstrate that the model that was proposed is robust and trustworthy when it comes to the classification of skin *** addition,the utilization of Explainable AI techniques,such as Grad-CAM visualizations,assists in highlighting the most significant lesion areas that have an impact on the decisions that are made by the model.

关键词： Skin lesions vision transformer CNN Xception deep learning network fusion explainable AI Grad-CAM skin cancer detection

来源：评论

学校读者我要写书评

暂无评论

ClipArtGAN: An Application of Pix2Pix Generative Adversarial Network for Clip Art Generation

引用

Multimedia Tools and Applications 2025年 1-25页

作者： Elnabawy, Reham H. Abdennadher, Slim Hellwich, Olaf Eldawlatly, Seif Technische Universität Berlin Berlin Germany Computer Science and Engineering Department Faculty of Media Engineering and Technology German University in Cairo Cairo Egypt Computer Science Department Faculty of Informatics and Computer Science German International University New Administrative Capital Egypt Computer and Systems Engineering Department Faculty of Engineering Ain Shams University 1 El-sarayat st. Abbassia Cairo Egypt Computer Science and Engineering Department The American University in Cairo Cairo Egypt School of Information Technology New Giza University Giza Egypt

An image can convey a thousand words. This statement emphasizes the importance of illustrating ideas visually rather than writing them down. Although detailed image representation is typically instructive, there are several contexts where simplification of image representations may be more informative. One way to provide an abstract representation of an object is via clip art, which is a simple graphic illustration of an object that can be included in documents or presentations. Clip art representation can be utilized in graphic design for concepts expression and daily life objects illustration. Furthermore, clip art had shown remarkable advantages when compared to real images in terms of editing and manipulation since clip art geometry can be easily edited in addition to the feasibility of storing it in a resolution-independent manner. However, it can be difficult to automatically create clip arts that resemble any object. Here, we propose a novel deep learning model, named ClipArtGAN, that utilizes Generative Adversarial Networks (GANs) to convert a photo of an object to its corresponding clip art. To our knowledge, this is the first GAN model based on Pix2Pix architecture that is capable of automatically generating clip art for any arbitrary object. Three training datasets were collected to train and compare two GAN models based on Pix2Pix and CycleGAN architectures. Our results demonstrate that the mean accuracy obtained using ClipArtGAN utilizing Pix2Pix outperforms that of CycleGAN by 58% and that of Google Images-based search by 54%. These results indicate the efficacy of ClipArtGAN in generating representative clip art images. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.

关键词： Generative adversarial networks

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：