检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Drummond, Ross Guiver, Chris Turner, Matthew C. Dept. of Automatic Control and Systems Engineering University of Sheffield SheffieldS1 3JD United Kingdom School of Computing Engineering & the Built Environment Edinburgh Napier University EdinburghEH10 5DT United Kingdom School of Electronics and Computer Science University of Southampton SouthamptonSO17 1BJ United Kingdom

With neural networks being used to control safety-critical systems, they increasingly have to be both accurate (in the sense of matching inputs to outputs) and robust. However, these two properties are often at odds with each other and a trade-off has to be navigated. To address this issue, this paper proposes a method to generate an approximation of a neural network which is certifiably more robust. Crucially, the method is fully convex and posed as a semi-definite programme. An application to robustifying model predictive control is used to demonstrate the results. The aim of this work is to introduce a method to navigate the neural network robustness/accuracy trade-off. © 2024, CC BY.

关键词： Model predictive control

来源：评论

学校读者我要写书评

暂无评论

Experimental Analysis of U-Net and Mask R-CNN for Segmentation of Synthetic Liquid Spray

Experimental Analysis of U-Net and Mask R-CNN for Segmentati...

引用

computing (ICOCO), IEEE International Conference on

作者： Refat Khan Pathan Wei Lun Lim Sian Lun Lau Chiung Ching Ho Prashant Khare Rahul Babu Koneru Dept. of Computing and Information Systems School of Engineering and Technology Sunway University Bandar Sunway Malaysia Dept. of Engineering School of Engineering and Technology Sunway University Bandar Sunway Malaysia Sunway DataXSight Research Cluster Sunway University Bandar Sunway Malaysia Department of Aerospace Engineering University of Cincinnati Cincinnati OH Department of Aerospace Engineering University of Maryland College Park Maryland USA

ISBN: (纸本)9781665489973

In digital image processing, segmentation is a process by which we can partition an image based on some variables to extract necessary elements. Unlike typical objects, it is complicated to segment dynamic objects from a synthetic fluid dataset where properties like position and shape change over time. Experiments on image segmentation over this dataset are conducted using U-Net (semantic segmentation) and Mask R-CNN (instance segmentation) to compare their results. The training dataset is generated from seven labelled images through data augmentation. Training on 1000 and validating on 200 images, Mask R-CNN achieved more epochs quickly. Around 1000 epochs for Mask R-CNN and 500 epochs for U-Net, both models reached a similar result in terms of F1 score and can segment the object in the new images.

关键词： Training Image segmentation Liquids Shape Semantic segmentation Fluid dynamics Digital images

来源：评论

学校读者我要写书评

暂无评论

A Dementia Diagnosis Technique Based on AI and Hardware Acceleration

A Dementia Diagnosis Technique Based on AI and Hardware Acce...

引用

Artificial Intelligence, Blockchain, and Internet of Things (AIBThings), IEEE International Conference on

作者： Kasem Khalil Mohammad M. R. Khan Mamun Ahmed Sherif Mohamed Elsersy Ahmad Abdel-Aliem Imam Kamal Abouzaid Electrical and Computer Engineering Department University of Mississippi MS USA Electrical and Computer Engineering Department Tennessee Technological University TN USA School of Computing Sciences and Computer Engineering University of Southern Mississippi MS USA Computer and Information Systems Dept. Higher Colleges of Technology AlAin Abu Dhabi UAE College of Osteopathic Medicine William Carey University Mississippi MS USA

The effects of Alzheimer’s disease (AD) are devastating, both personally and within the patient’s family, as the disease progresses slowly over many years. It could significantly affect illness consequences and outcomes if detected early and accurately. Non-invasive medical testing often uses blood bio-samples because they are easy to collect and relatively cheap to analyze. This paper proposes a federated learning (FL)-based diagnostic model for AD using blood bio-samples. We have used blood bio-sample data sets downloaded from the ADNI website to test and compare the efficacy of our models. According to the enormous data collected for early AD detection, we used a hardware acceleration scheme to implement our FL model to fasten the training and testing operations. The hardware-accelerator method is implemented using VHDL and Altera 10 GX FPGA. The simulation results show the proposed algorithms reach an accuracy and sensitivity of 89% and 87%, respectively, for early detection while taking less time to train than other state-of-the-art algorithms. The proposed algorithms also consume 52-172 mW, which makes them suitable for constrained devices.

关键词：

来源：评论

学校读者我要写书评

暂无评论

ASD Detection using Higuchi’s Fractal Dimension from EEG

ASD Detection using Higuchi’s Fractal Dimension from EEG

引用

IEEE International Conference on Biomedical engineering, Computer and Information Technology for Health

作者： Zahrul Jannat Peya Md. Ferdous M. A. H. Akhand Mohammed Golam Zilani Nazmul Siddique Dept. of Computer Science and Engineering Khulna University of Engineering & Technology Khulna Bangladesh James Cook University Queensland Australia School of Computing Engineering and Intelligent Systems Ulster University United Kingdom

ISBN: (数字)9781665466219

ISBN: (纸本)9781665466226

Autism or Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder marked by repetitive and characteristic patterns of behavior as well as social communication and interaction impairments. Nowadays ASD is a great concern worldwide and its detection is an important issue for the better treatment. As ASD is neurodevelopmental disorder, brain signals play, especially electroencephalography (EEG), is shown potential sources for ASD detection. There are different approaches for ASD detection with processing and/or transforming EEG signals. This study investigated ASD detection employing fractal dimension measure on EEG data. Higuchi’s Fractal Dimension (HFD) is measured in resting-state eyes-closed EEG recording of 25 subjects. It is identified that HFD is sensitive to the brain activity and ASD detection is possible from HFD values.

关键词： Autism Brain Electric potential Psychology Electroencephalography Fractals Complexity theory

来源：评论

学校读者我要写书评

暂无评论

Comparison of CNN visualization methods to aid model interpretability for detecting alzheimer’s disease

Comparison of CNN visualization methods to aid model interpr...

引用

International workshop on Algorithmen - Systeme - Anwendungen, 2020

作者： Dyrba, Martin Pallath, Arjun H. Marzban, Eman N. Rostock Germany Institute of Visual & Analytic Computing University of Rostock Rostock Germany University Medical Center Rostock Rostock Germany Biomedical Engineering and Systems Dept. Faculty of Engineering Cairo University Giza Egypt

ISBN: (纸本)9783658292669

Advances in medical imaging and convolutional neural networks (CNNs) have made it possible to achieve excellent diagnostic accuracy from CNNs comparable to human raters. However, CNNs are still not implemented in medical trials as they appear as a black box system and their inner workings cannot be properly explained. Therefore, it is essential to assess CNN relevance maps, which highlight regions that primarily contribute to the prediction. This study focuses on the comparison of algorithms for generating heatmaps to visually explain the learned patterns of Alzheimer’s disease (AD) classification. T1-weighted volumetric MRI data were entered into a 3D CNN. Heatmaps were then generated for different visualization methods using the iNNvestigate and keras-vis libraries. The model reached an area under the curve of 0.93 and 0.75 for separating AD dementia patients from controls and patients with amnestic mild cognitive impairment from controls, respectively. Visualizations for the methods deep Taylor decomposition and layer-wise relevance propagation (LRP) showed most reasonable results for individual patients matching expected brain regions. Other methods, such as Grad-CAM and guided backpropagation showed more scattered activations or random areas. For clinically research, deep Taylor decomposition and LRP showed most valuable network activation patterns. © Springer Fachmedien Wiesbaden GmbH, ein Teil von Springer Nature 2020.

关键词： Diagnosis

来源：评论

学校读者我要写书评

暂无评论

Federated Learning Model for Early Detection of Dementia Using Blood Biosamples

Federated Learning Model for Early Detection of Dementia Usi...

引用

Artificial Intelligence, Blockchain, and Internet of Things (AIBThings), IEEE International Conference on

作者： Mohamed Elsersy Ahmed Sherif Ahmad Abdel-Aliem Imam Mohammad M. R. Khan Mamun Kasem Khalil Mohamed Haitham Computer and Information Systems Dept. Higher Colleges of Technology Abu Dhabi UAE School of Computing Sciences and Computer Engineering University of Southern Mississippi MS USA College of Osteopathic Medicine William Carey University Mississippi MS USA Electrical and Computer Engineering Department Tennessee Technological University TN USA Electrical and Computer Engineering Department University of Mississippi MS USA University of Prince Edward Island (UPEI) Cairo Egypt

Alzheimer’s disease (AD) is a serious, long-term health problem that causes much pain and loss for the person with it and their family. Its early and accurate detection might result in a substantial reduction of the disease outcomes and consequences. Blood biosamples are a simple and inexpensive technique in medical testing. This paper proposes diagnostic models for blood biosamples based on federated learning (FL) and its modifications to detect AD early. Our experiments used blood biosample data sets from the ADNI website to evaluate our models. Our performance analysis indicates that our algorithms are more accurate and achieve an accuracy of 87% for early detection.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Data-driven Distributed and Localized Model Predictive Control

arXiv

引用

arXiv 2021年

作者： Alonso, Carmen Amo Yang, Fengjun Matni, Nikolai Dept. of Computing and Mathematical Sciences California Institute of Technology PasadenaCA United States Dept. of Computer and Information Sciences University of Pennsylvania PhiladelphiaPA United States Dept. of Electrical and Systems Engineering University of Pennsylvania PhiladelphiaPA United States

Motivated by large-scale but computationally constrained settings, e.g., the Internet of Things, we present a novel data-driven distributed control algorithm that is synthesized directly from trajectory data. Our method, data-driven Distributed and Localized Model Predictive Control (D3LMPC), builds upon the data-driven System Level Synthesis (SLS) framework, which allows one to parameterize closed-loop system responses directly from collected open-loop trajectories. The resulting model-predictive controller can be implemented with distributed computation and only local information sharing. By imposing locality constraints on the system response, we show that the amount of data needed for our synthesis problem is independent of the size of the global system. Moreover, we show that our algorithm enjoys theoretical guarantees for recursive feasibility and asymptotic stability. Finally, we also demonstrate the optimality and scalability of our algorithm in a simulation experiment. © 2021, CC BY.

关键词： Model predictive control

来源：评论

学校读者我要写书评

暂无评论

Approximate novelty search

arXiv

引用

arXiv 2021年

作者： Singh, Anubhav Lipovetzky, Nir Ramirez, Miquel Segovia-Aguas, Javier School of Computing and Information Systems University of Melbourne Australia Electrical and Electronic Engineering University of Melbourne Australia Dept. Information and Communication Technologies Universitat Pompeu Fabra Spain

Width-based search algorithms seek plans by prioritizing states according to a suitably defined measure of novelty, that maps states into a set of novelty categories. Space and time complexity to evaluate state novelty is known to be exponential on the cardinality of the set. We present novel methods to obtain polynomial approximations of novelty and width-based search. First, we approximate novelty computation via random sampling and Bloom filters, reducing the runtime and memory footprint. Second, we approximate the best-first search using an adaptive policy that decides whether to forgo the expansion of nodes in the open list. These two techniques are integrated into existing width-based algorithms, resulting in new planners that perform significantly better than other state-of-the-art planners over benchmarks from the International Planning Competitions. © 2021, CC BY.

关键词： Polynomial approximation

来源：评论

学校读者我要写书评

暂无评论

ROBOSTEAMSEN Project - Training SEN Teachers to Use Robotics for Fostering STEAM and Develop Computational Thinking*

ROBOSTEAMSEN Project - Training SEN Teachers to Use Robotics...

引用

International Symposium on Computers in Education (SIIE)

作者： Miguel Á. Conde Francisco J. Rodríguez-Sedano Francisco J. García-Peñalvo Jose Gonçalves Ilkka Jormanainen Andrea Anzanello Jonny Filipe Ribeiro Alves Raquel Fernández Hernáandez Alina- Andreia Ailincai Dept. of Mechanics Computer Science and Aerospace Engineering Universidad de Leon Leon Spain Dept. Electrical Systems and Automation Engineering Universidad de Leon Leon Spain Dpto. de Informatica y Autormatica - IUCE Universidad de Salamanca Salamanca Espana Research Centre in Digitalization and Intelligent Robotics (CeDRI) Instituto Politecnico de Braganca Braganca Portugal School of Computing University of Eastern Finland Joensuu Finland Dept of Project Management Pixel Florence Italy Colegio Internato Dos Carvalhos Carvalhos Portugal Colegio de Educacion Especial Ntra. Sra. del Sagrado Corazon de Leon Leon Spain EuroEd Primary School Lasi Romania

ISBN: (数字)9798350376616

ISBN: (纸本)9798350376623

Our contemporary society necessitates professionals equipped with 21st-century skills. Disciplines within Science, Technology, engineering, Arts, and Mathematics (known as STEAM) have been particularly effective in fostering these skills. However, when considering students with disabilities, especially those with intellectual or developmental disabilities (IDD), this assertion often falls short. In this context, the RoboSTEAMSEN project emerges as an initiative designed to enhance educational processes by providing teachers of IDD students with the necessary resources to promote STEAM engagement. The project proposes the use of active learning methodologies and robotics to achieve this goal. The primary objective of the project is realized through several strategies: understanding the needs of students with disabilities and adapting the use of robotics and active learning methodologies accordingly; training teachers in the use of these resources; and creating a platform to exchange experiences, resources, lessons learned, tools, case scenarios, etc., while reaching other potential stakeholders such as caregivers and policymakers. The main outcomes of the project are teacher training programs and the development of associated competencies, tools to identify and classify resources for the students, and technological platforms to ensure the sustain ability of the project once it concludes.

关键词： Training Computers Ecosystems Collaboration Fasteners People with disabilities Mathematics

来源：评论

学校读者我要写书评

暂无评论

Backward Radiation Reduction and Bandwidth Enhancement of Metamaterial Antenna for UWB Applications

Backward Radiation Reduction and Bandwidth Enhancement of Me...

引用

2021 Emerging Technology in computing, Communication and Electronics, ETCCE 2021

作者： Uddin, M. Jasim Ullah, M. Habib Bentley, Barry L. Shakib, M.N. Islam, Syed Zahurul Rahman, Md Arafatur Cardiff Metropolitan University Cardiff School of Technologies Department of Applied Computing and Engineering United Kingdom University of Malaya Faculty of Engineering Department of Electrical Engineering Kuala Lumpur50603 Malaysia Cardiff Metropolitan University Cardiff School of Technologies Department of Computer Science United Kingdom Green University of Bangladesh Faculty of Science and Engineering Dept. of Electrical and Electronics Engineering Dhaka1207 Bangladesh University Tun Hussein Onn Malaysia Faculty of Electrical and Electronic Engineering Power Integration Systems Johor Malaysia Faculty of Science and Engineering University of Wolverhampton United Kingdom

ISBN: (纸本)9781665483643

A metamaterial inspired planar-patterned microstrip patch antenna, feeding on a Duroid substrate inset bottom ground soft-surface, is presented for ultra-wide band (UWB) applications. The present antenna is configured on a step-edge periodic unit-cell pattern, with a tiny air gap etched on the metal patch and finite ground plane. By employing these structures an ultra-wide band from 4 GHz ~ 16.2 GHz is achieved. The important advantage is retaining the suppression of back radiation while removing grounded edge metallic parts and replacing them with the finite cut-off square unit cell. The feeding ground plane and radiating patch are specially configured to develop artificial structures using series capacitance and shunt inductance. The efficiency of the proposed antenna is above 99.1%. To enhance gain and bandwidth, this antenna has etched Jerusalem crossed slots on the ground plane and star-slots on the patch, with the enhanced gain reaching a maximum of 8 dBi at 9.2 GHz. The antenna's overall dimensions exhibit 0.72λ × 0.83λ × 0.05λ at a 10.1 GHz cut-off frequency. The high absorption gain, directivity, and enhanced frequency bandwidth, verified by numerical modeling and experimentation, ensure the proposed antenna is well suited to any UWB application. © 2021 IEEE.

关键词： Metamaterial antennas

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：