检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Weiss, Jonah Obrien Alves, Tiago Kundu, Sandip Dept. of Electrical and Computer Engineering University of Massachusetts Amherst AmherstMA United States Dept. of Informatics and Computer Science State University of Rio de Janeiro - UERJ Rio de Janeiro Brazil

The prevalence and success of Deep Neural Network (DNN) applications in recent years have motivated research on DNN compression, such as pruning and quantization. These techniques accelerate model inference, reduce power consumption, and reduce the size and complexity of the hardware necessary to run DNNs, all with little to no loss in accuracy. However, since DNNs are vulnerable to adversarial inputs, it is important to consider the relationship between compression and adversarial robustness. In this work, we investigate the adversarial robustness of models produced by several irregular pruning schemes and by 8-bit quantization. Additionally, while conventional pruning removes the least important parameters in a DNN, we investigate the effect of an unconventional pruning method: removing the most important model parameters based on the gradient on adversarial inputs. We call this method Greedy Adversarial Pruning (GAP) and we find that this pruning method results in models that are resistant to transfer attacks from their uncompressed counterparts. Code is available at [1]. © 2022, CC BY.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Biological age estimation from ECG using Artificial Intelligence approaches

Biological age estimation from ECG using Artificial Intellig...

引用

biomedical Engineering Conference (JIBEC), Jordanian International

作者： Jana Sleiman Mohamad Daher Mohamad Khalil Ahmad Diab Abdallah Kassem Computer and Communications Engineering Lebanese University Tripoli Lebanon Dep. of Computer Science/Dep. of Computer Science & Information Technology Faculty of Science/Faculty of Applied Sciences & Technology Tripoli Lebanon Electrical Engineering Lebanese University Tripoli Lebanon Biomedical Engineering Lebanese International University Beirut Lebanon ECCE Dept. Faculty Of Engineering Notre Dame University Zouk Mosbeh Lebanon

ISBN: (数字)9798331539542

ISBN: (纸本)9798331539559

This paper explores the application of Artificial Intelligence (AI) techniques in healthcare, specifically focusing on electrocardiogram (ECG) data analysis and biological age estimation. The study begins with an overview of ECG as a diagnostic tool and discusses the integration of Deep Learning (DL) models, particularly Convolutional Neural Networks (CNNs), in healthcare for various applications. Previous studies reviewed in this context consistently identify CNNs as the superior model for ECG analysis and prediction tasks. The primary research investigates the use of Machine Learning (ML) algorithms to estimate biological age from ECG data, employing metrics such as Mean Absolute Error (MAE) and variance to evaluate model performance. Subsequently, DL models are implemented and compared, demonstrating superior accuracy in estimating the biological age of new patients. While both ML and DL approaches were initially explored, this paper primarily focuses on deep learning methodologies due to their demonstrated superiority in healthcare applications. The findings highlight the efficacy of deep learning over traditional ML methods, underscoring its potential for enhancing diagnostic accuracy and personalized patient care. For instance, the most accurate deep learning model, the Convolutional Neural Network (CNN), achieved a Mean Absolute Error (MAE) of 1.49 and an R 2 score of 0.53, whereas the best-performing traditional ML model, Random Forest, achieved an MAE of 1.54 and an R 2 score of 0.55. These results clearly indicate the advantage of deep learning models in capturing complex patterns in ECG data, leading to more precise and reliable estimations of biological age.

关键词： Deep learning Maximum likelihood estimation Accuracy Data analysis Biological system modeling Medical services Electrocardiography Data models Convolutional neural networks Random forests

来源：评论

学校读者我要写书评

暂无评论

Design of photoacoustic modular helmet for neonatatal neurocritical care

Design of photoacoustic modular helmet for neonatatal neuroc...

引用

IEEE International Symposium on Applications of Ferroelectrics (ISAF)

作者： Ananya Tandri Keshuai Xu May W. Chen Emad Boctor Jeeun Kang Dept. Biomedical Engineering Johns Hopkins University Baltimore MD USA Dept. Computer Science Johns Hopkins University Baltimore MD USA Dept. Pediatrics Johns Hopkins University Baltimore MD USA Lab. for Computational Sensing & Robotics Johns Hopkins University Baltimore MD USA Dept. Anesthesiology and Critical Care Medicine Johns Hopkins University Baltimore MD USA

ISBN: (数字)9798350371901

ISBN: (纸本)9798350371918

Perinatal arterial ischemic stroke (PAIS) is a form of focal vascular neurological injury that occurs during or shortly after birth. We recently proposed a modular PA helmet (PAH) system, in which multiple modules constitutes a flexible helmet to accommodate any neonatal head shape and size to provide volumetric monitoring of neonatal brain oxygen saturation. Here, optical tracking of the PAH configuration enables spatial image reconstruction. Given the large freedom in PAH configuration, it is critical to understand the implications of various design parameters in imaging quality. In this paper, we present our simulation-aided design approach using acoustic and optical simulation tools. Our design criteria are to optimize 1) signal contrast and 2) spatial resolution with different number of modules and their mean distances, and is further challenged with modular position error inevitable in modular position recognition. Results of the simulation established an acceptable error threshold for optical tracking and informed module prototype configuration and shape. The prototype based on the optimal design specifications is also demonstrated.

关键词： Pediatrics Head Shape Prototypes Optical saturation Optical variables control Optical imaging Adaptive optics Acoustics Safety

来源：评论

学校读者我要写书评

暂无评论

Real-Time Automatic Checkout via Prompt-Based Product Extraction and Cross-Domain Learning

Real-Time Automatic Checkout via Prompt-Based Product Extrac...

引用

International Conference on Machine Learning and Applications (ICMLA)

作者： Tobias Pettersson Maria Riveiro Tuwe Löfström University of Skövde Jönköping University ITAB Shop Products AB Jönköping Sweden Dept. of Computer Science and Informatics Jönköping University Jönköping Sweden Dept. of Computing Jönköping University Jönköping Sweden

ISBN: (数字)9798350374889

ISBN: (纸本)9798350374896

Automatic checkout systems are designed to predict a complete shopping receipt using an image from the checkout area. These systems require high classification accuracy across numerous classes and must operate in real-time, despite domain differences between training data and real-world conditions. Building on recent advancements, we propose a method that outperforms current solutions and can be applied in real-time in automatic checkout systems. Our method leverages the Segment Anything Model to extract high-quality masks from lab product images, which are then transformed into synthetic checkout images and adapted to the real domain using contrastive unpaired translation. We train a product recognition model with data augmentation, named SCA+Y8, and further improve it through fine-tuning with pseudo-labels from unlabeled checkout images, resulting in an improved model called SCAFT+Y8. SCAFT+Y8 achieves a great increase in state-of-the-art performance, with an average receipt classification accuracy of 97.58%, and shows strong performance in smaller models, indicating the potential for deployment on low-cost edge devices.

关键词： Performance evaluation Adaptation models Accuracy Translation Image edge detection Training data Machine learning Data augmentation Data models Real-time systems

来源：评论

学校读者我要写书评

暂无评论

Non-collocated Vibration Suppression using Delay-Based Controllers with Multiple Fixed Delays*

引用

IFAC-PapersOnLine 2023年第2期56卷 180-185页

作者： Adrian Saldanha Wim Michiels Tomáš Vyhlídal Department of Computer Science KU Leuven Celestijnenlaan 200A 3001 Leuven Belgium Dept. Instrumentation and Control Engineering Faculty of Mechanical Engineering and Czech Institute of Informatics Robotics and Cybernetics Czech Technical University in Prague Prague 6 Czechia

The problem of non-collocated vibration absorption is targeted utilizing a multiparameter delay-based controller. The combined objective is to achieve closed-loop stability and vibration suppression at a target location, which is different from the position of the absorber. This problem can be translated into an optimization problem of shaping the poles of a closed-loop system with delays subject to zero-location constraints. The solution approach involves remodelling the closed-loop system as a system of delay-differential algebraic equations (DDAEs) which allows to systematically account for delays in the control law, thereby enabling a straightforward technique for computation of controller parameters. Finally, the proposed design method is validated by simulation on a lumped parameter model of a mass-spring-damper system.

关键词： vibration suppression time-delay systems DDAE optimization

来源：评论

学校读者我要写书评

暂无评论

A New Approach for Spread Spectrum Steganography in Audio Containers

A New Approach for Spread Spectrum Steganography in Audio Co...

引用

International Conference on Control Systems, Mathematical Modelling, Automation and Energy Efficiency (SUMMA)

作者： Stanimir Zhelezov Krasimir Kordov Atanas Nachev Daniela Pavlova Panayot Gindev Nikolay Litchkov Konstantin Nesterov dept. Computer Science and Technologies Konstantin Preslavsky University of Shumen Shumen Bulgaria dept. Computer Informatics Konstantin Preslavsky University of Shumen Shumen Bulgaria University of Library Studies and Information Technologies Sofia Bulgaria Institute of Robotics at Bulgarian Academy of Sciences Sofia Bulgaria Institute of Metal Science Equipment and Technologies with Center for Hydro- and Aerodynamics Acad. A. Balevski Bulgarian Academy of Sciences Sofia Bulgaria

ISBN: (数字)9798331532178

ISBN: (纸本)9798331532185

The most common approach among researchers is to implement cryptographic techniques and methods to provide a secure and reliable way to transmit sensitive information. This paper proposes a steganographic embedding algorithm based on Circle Map and Modified Rotation Equations. Audio files randomly selected from the Internet are used for the purpose of the study. Messages of different sizes were embedded in them using the proposed steganographic embedding algorithm. The accumulated database of more than 200 containers and stegofiles was analyzed using the most common steganalysis methods. From the analyses, it can be clearly seen that the proposed method of steganographic embedding in audio containers provides a very high level of security in the transmission of hidden information.

关键词： Steganography Automation Databases Containers Control systems Mathematical models Generators Energy efficiency Internet Reliability

来源：评论

学校读者我要写书评

暂无评论

A Software Package for Analyzing and Evaluating Children with Autism Spectrum Disorder (ASD) in Sri Lanka

A Software Package for Analyzing and Evaluating Children wit...

引用

2020 International Conference on Decision Aid sciences and Application, DASA 2020

作者： Nazardeen, Nuska Yapa, Lmesha Imthath, Mohamed Pathirana, Suneth Jayathunga, Poornnima Uva Wellassa University Dept. Computer Science and Informatics Badulla Sri Lanka

ISBN: (纸本)9781728196770

In the fast-moving world of today, the recent statistics of population patron, 1 in 63 children are affected with Autism. Up-to-date no well-defined treatment for Autism Spectrum Disorder (ASD) has been invented yet. In this study, an easy, cost-effective, time saving and attractive software package is introduced. The main objective of developing such a package is to differentiate neurotypical and ASD children's behavior patterns. This package was designed by considering three main impaired areas of ASD which are Eye Contact, Maturity Level, and Intelligence Level. Therefore, the developed software package is comprised of an Eye Movement Tracking tool, Maturity Level Analyzer, and Intelligence Level Measuring tool. In the Eye Movement Tracking tool, a common sample video can be shown to the participants (Neurotypical Children and ASD Children) and their eye movements shall be recorded. Then the recorded data will be processed using symmetric mass center algorithm and displayed the results in three different graphs. Maturity Level Analyzer provides a drawing canvas where participants shall be allowed to draw a given shape. Then, the analysis of the drawing shall be done by comparing the standard deviation (SD) of participants' data. Intelligence level Measuring Tool compromised with color and number-based activities where participants' responses will be taken to decide the child's condition. Testing and evaluation of the system were done with two different types of participants as three (3) ASD patients and ten (10) neurotypical persons from the age groups of 3-5 years. To evaluate the accuracy and performance of the tools, two tests as 'pre-test' and 'post-test' were done with the participants. The tested results have shown a significant difference between ASD patients and neurotypical children's behavior. Thus, this set of tools is more effective for taking decisions in ASD prediction and measuring the improvements of ASD children with treatments. © 2020 IEEE.

关键词： Population statistics

来源：评论

学校读者我要写书评

暂无评论

Optimizing NFC-Based Wearable Sensors for Arterial Pulse Monitoring: A Comparative Study of Sampling Rates and Machine Learning Models

Optimizing NFC-Based Wearable Sensors for Arterial Pulse Mon...

引用

International Workshop on Wearable and Implantable Body Sensor Networks (BSN)

作者： Anita Ghandehari Jorge Alfonso Tavares-Negrete Xiaochang Pei Jerome Rajendran Shingirirai Chakoma Rahim Esfandyarpour Dept. of Electrical Engineering and Computer Science Integrated Nano BioElectronics Innovation Lab The Henry Samueli School of Engineering University of California Irvine Irvine USA Dept. of Biomedical Engineering Integrated Nano BioElectronics Innovation Lab The Henry Samueli School of Engineering University of California Irvine Irvine USA

ISBN: (数字)9798331530143

ISBN: (纸本)9798331530150

This paper details the development and optimization of a novel NFC-based wearable pressure sensor designed for arterial pulse monitoring with a focus on energy efficiency and data transmission management. By integrating this sensor with various machine learning models for human activity recognition using arterial pulse signals, we evaluate their impact on optimizing the sampling rate. This optimization directly influences power consumption and data transmission rates. A comparative analysis reveals that the 1-dimensional Convolutional Neural Network (1D-CNN) maintains an accuracy of over 92% while allowing the sampling rate to be lowered from 100 Hz to 40 Hz. This reduction leads to a 35% decrease in power consumption and a 50% decrease in data transmission needs. In contrast, traditional models such as Random Forest (RF) and Support Vector Machine (SVM) require a minimum of 70 Hz to maintain comparable accuracy, achieving only a 19% reduction in power and a 30% reduction in data transmission rates. Our findings emphasize the critical role of model selection in optimizing the arterial pulse sampling rate, enhancing the energy efficiency and operational effectiveness of our wearable health monitoring device.

关键词： Support vector machines Radio frequency Pressure sensors Power demand Accuracy Energy efficiency Data communication biomedical monitoring Monitoring Optimization

来源：评论

学校读者我要写书评

暂无评论

Enhancing Early Diagnosis of Autism With Machine Learning Algorithms Using Postural Control Features

Enhancing Early Diagnosis of Autism With Machine Learning Al...

引用

IEEE Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON)

作者： Maya Nachega Sugru A. Yahaya Ila Wallace Ting Liu Damian Valles Rachel M. Koldenhoven Yumen Li Dept. of Biomedical Engineering Vanderbilt University Nashville USA Weldon School of Biomedical Engineering Purdue University West Lafayette USA Dept. of Computer Science Texas State University San Marcos USA College of Graduate Studies Texas A&M University-San Antonio San Antonio USA Ingram School of Engineering Texas State University San Marcos USA Dept. of Health & Human Performance Texas State University San Marcos USA

ISBN: (数字)9798331540906

ISBN: (纸本)9798331540913

Autism Spectrum Disorder (ASD) is a common neurodevelopmental disorder whose biological cause is still not well understood. Due to this, as well as the gradual onset patterns of the disorder’s symptoms, early identification and diagnosis of ASD proves to be challenging. This study aimed to provide a potential early diagnostic tool for ASD by training machine learning classifiers with linear and nonlinear postural control data features. For the research methodology, center of pressure (COP) data for 38 children ages 5-16 was collected using a force plate during both eyes opened and eyes closed conditions. Each trial was 20 seconds, and three trials were completed for each condition. The raw data was preprocessed, and six COP variables were calculated for each trial: anteroposterior (AP) displacement, mediolateral (ML) displacement, elliptical sway area, COP travel distance, AP multiscale entropy complexity, and finally, ML multiscale entropy complexity. After preprocessing, these data features were trained to seven machine learning models to classify participants as having ASD or typical development (TD). Our experiment showed that all seven machine learning models could appropriately classify participants with an accuracy of $79 \%$ and above. The best classification performance had an accuracy of almost $97 \%$ and was done by the model that utilized the random forest algorithm for classification. Additionally, the data feature that was the most important across machine learning algorithms was AP complexity. These results have strengthened the validity of a machine learning approach for early ASD diagnosis. In future research, more data should be collected beyond the 38 participants obtained in this study, or data augmentation should be done to generate more data to train the models. Future research should also put efforts into obtaining more robust and representative datasets, exploring best multiscale entropy parameter values, and exploring why certain data featu

关键词： Training Autism Machine learning algorithms Accuracy Machine learning Data augmentation Entropy Data models Complexity theory Random forests

来源：评论

学校读者我要写书评

暂无评论

GenNER - A highly scalable and optimal NER method for text-based gene and protein recognition

GenNER - A highly scalable and optimal NER method for text-b...

引用

2021 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2021

作者： Fotseu, Ernest Basile Fotseu Nembot, Thierry Kongne Sani, Rajesh K. Gadhamshetty, Venkataramana Etienne Gnimpieba, Z. Bomgni, Alain Bertrand University of Dschang Dept. of Computer Science Dschang Cameroon Dept. of Chemical and Biological Engineering Rapid city United States South Dakota School of Mines and Technology Dept. of Civil and Environmental Engineering Rapid city United States University of South Dakota Dept. of Biomedical Engineering Vermillion United States

ISBN: (纸本)9781665401265

Nowadays, there are a large number of models in the scientific literature capable of recognizing and extracting gene mentions from a given text. Several data sets have been developed to facilitate the learning process of these models. However, very few models are able to increase their knowledge and performance progressively from new annotated text but also to take into account the granularity of the input text of the model. Our proposed solution, GenNER, is a method for recognizing gene/protein mentions from free text. GenNER relies on continuous learning and a text granularization algorithm as input to the model, which allows it to achieve better performance. Its evaluation process was done around BioCreative II annotated datasets;we obtained an average F1-score of 0.9704, which outperforms current methods. © 2021 IEEE.

关键词： Proteins

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：