检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Liao, Yiwen Rivoir, Jochen Latty, Raphaël Yang, Bin Institute of Signal Processing and System Theory University of Stuttgart Pfaffenwaldring 47 Baden-Wuerttemberg Stuttgart70569 Germany Applied Research and Venture Team Advantest Europe GmbH Herrenberger Strasse 130 Baden-Wuerttemberg Boeblingen71034 Germany

Feature selection has drawn much attention over the last decades in machine learning because it can reduce data dimensionality while maintaining the original physical meaning of features, which enables better interpretability than feature extraction. However, most existing feature selection approaches, especially deep-learning-based, often focus on the features with great importance scores only but neglect those with less importance scores during training as well as the order of important candidate features. This can be risky since some important and relevant features might be unfortunately ignored during training, leading to suboptimal solutions or misleading selections. In our work, we deal with feature selection by exploiting the features with less importance scores and propose a feature selection framework based on a novel complementary feature mask. Our method is generic and can be easily integrated into existing deep-learning-based feature selection approaches to improve their performance as well. Experiments have been conducted on benchmarking datasets and shown that the proposed method can select more representative and informative features than the state of the art. Copyright © 2022, The Authors. All rights reserved.

关键词： Feature Selection

来源：评论

学校读者我要写书评

暂无评论

Low Noise Inverse Magnetoelectric Magnetic Field Sensor

arXiv

引用

arXiv 2023年

作者： Thormählen, Lars Hayes, P. Elzenheimer, E. Spetzler, E. McCord, J. Schmidt, G. Höft, M. Meyners, D. Quandt, E. Inorganic Functional Materials Institute for Materials Science Kiel University Kiel24143 Germany Digital Signal Processing and System Theory Institute of Electrical Engineering and Information Technology Kiel University Kiel24143 Germany Microwave Engineering Institute of Electrical Engineering and Information Technology Kiel University Kiel24143 Germany Nanoscale Magnetic Materials Institute for Materials Science Kiel University Kiel24143 Germany

In the development of any type of magnetic field sensor based on magnetic films, special consideration must be given to the magnetic layer component. The presented work investigates the use of flux closing magnetostrictive multilayers for inverse magnetoelectric sensors. In such a type of magnetic field sensor, highly sensitive AC and DC field detection relies on strong excitation of the incorporated magnetic layers by piezoelectrically driven cantilever oscillation at mechanical resonances. The provoked periodic flux change is influenced by the magnetic field to be measured and is picked up by a coil, which generates the measured output. The effect of the magnetic multilayer on linearity, noise behavior, and detection limit of DC and AC signals is investigated. This study demonstrates the next step for inverse magnetoelectric thin film sensors, which achieve one order of magnitude improved detection limits with less than 8 pT/Hz1/2 at 10 Hz and 18 pT/Hz1/2 at DC using exchange bias stabilized magnetic multilayers for obtaining flux closure. Copyright © 2023, The Authors. All rights reserved.

关键词： Magnetic fields

来源：评论

学校读者我要写书评

暂无评论

Continual Learning Through One-Class Classification Using VAE

Continual Learning Through One-Class Classification Using VA...

引用

IEEE International Conference on Acoustics, Speech and signal processing

作者： Felix Wiewel Andreas Brendle Bin Yang Institute of Signal Processing and System Theory University of Stuttgart Germany

ISBN: (数字)9781509066315

ISBN: (纸本)9781509066322

Artificial neural networks (ANNs) suffer from catastrophic forgetting, a sharp decrease in performance on previously learned tasks, when trained on a new task without constant rehearsal. In this paper, we propose a new method for overcoming this phenomenon based on one-class classification. It is not only able to incrementally learn new but also detect unknown classes. This is a desirable property, since it enables the detection of new and unknown classes in a stream of data and adaption to a changing environment. Experiments on commonly used continual learning setups show competitive results and verify the concept.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Supervised Canonical Correlation Analysis of Data on Symmetric Positive Definite Manifolds by Riemannian Dimensionality Reduction

Supervised Canonical Correlation Analysis of Data on Symmetr...

引用

IEEE International Conference on Acoustics, Speech and signal processing

作者： Faezeh Fallah Bin Yang Institute of Signal Processing and System Theory University of Stuttgart Germany

ISBN: (数字)9781509066315

ISBN: (纸本)9781509066322

Most computer vision problems entail data that reside on Riemannian manifolds. Canonical correlation analysis (CCA) is a powerful method that captures correlations between any two sets of matrices. In this paper, we propose a framework for a supervised CCA of manifold-based data. This framework aims to find the optimal dimensionality reduction maps that maximize the discriminative power of any classifier in the reduced dimensional space and the correlation between the projected sets. This allows to incorporate the CCA into a classifier that analyzes multichannel or multimodal data on separate manifolds. The proposed method is evaluated on the challenging task of segmenting cardiac adipose tissues on fat-water (2-channel) magnetic resonance images.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A MIMO Radar-based Few-Shot Learning Approach for Human-ID

A MIMO Radar-based Few-Shot Learning Approach for Human-ID

引用

European signal processing Conference (EUSIPCO)

作者： Pascal Weller Fady Aziz Sherif Abdulatif Urs Schneider Marco F. Huber Department of Bio-mechatronic Systems Fraunhofer IPA Stuttgart Germany Institute of Signal Processing and System Theory University of Stuttgart Germany Institute for Industrial Manufacturing and Factory Operation IFF University of Stuttgart Germany Fraunhofer IPA Center for Cyber Cognitive Intelligence Stuttgart Germany

ISBN: (数字)9789082797091

ISBN: (纸本)9781665467995

Radar for deep learning-based human identification has become a research area of increasing interest. It has been shown that micro-Doppler $(\mu-\mathbf{D})$ can reflect the walking behavior, through capturing the periodic limbs micro-motions. One of the main aspects is maximizing the number of included classes, while considering the real-time and training dataset size constraints. In this paper, a multiple-input-multiple-output (MIMO) radar is used to formulate micro-motion spectrograms of the elevation angular velocity $(\mu-\omega)$ . The effectiveness of concatenating this newly-formulated spectrogram with the commonly used $\mu-\mathrm{D}$ ones is investigated. To accommodate for non-constrained real walking motion, an adaptive cycle segmentation framework is utilized and a metric learning network is trained on half gait cycles $(\approx 0.5\mathrm{s})$ . Studies on the effects of various numbers of classes (5–20), different dataset sizes, and varying observation time windows (1–2 s) are conducted. A non-constrained walking dataset of 22 subjects is collected with different aspect angles with respect to the radar. The proposed few-shot learning (FSL) approach achieves a classification error of 11.3 % with only 2 min of training data per subject.

关键词： Legged locomotion Training Three-dimensional displays Training data Radar Real-time systems Angular velocity

来源：评论

学校读者我要写书评

暂无评论

Organ-Based Chronological Age Estimation Based on 3D MRI Scans

Organ-Based Chronological Age Estimation Based on 3D MRI Sca...

引用

European signal processing Conference (EUSIPCO)

作者： Karim Armanious Sherif Abdulatif Anish Rao Bhaktharaguttu Thomas Küstner Tobias Hepp Sergios Gatidis Bin Yang University of Stuttgart Institute of Signal Processing and System Theory Stuttgart Germany University of Tübingen Tübingen Germany King’s College London London England

ISBN: (数字)9789082797053

ISBN: (纸本)9781728150017

Individuals age differently depending on a multitude of different factors such as lifestyle, medical history and genetics. Often, the global chronological age is not indicative of the true ageing process. An organ-based age estimation would yield a more accurate health state assessment. In this work, we propose a new deep learning architecture for organ-based age estimation based on magnetic resonance images (MRI). The proposed network is a 3D convolutional neural network (CNN) with increased depth and width made possible by the hybrid utilization of inception and fire modules. We apply the proposed framework for the tasks of brain and knee age estimation. Quantitative comparisons against concurrent MR-based regression networks and different 2D and 3D data feeding strategies illustrated the superior performance of the proposed work.

关键词： Three-dimensional displays Magnetic resonance imaging Two dimensional displays Estimation Computer architecture signal processing Task analysis

来源：评论

学校读者我要写书评

暂无评论

Uncertainty-Based Biological Age Estimation of Brain MRI Scans

Uncertainty-Based Biological Age Estimation of Brain MRI Sca...

引用

IEEE International Conference on Acoustics, Speech and signal processing

作者： Karim Armanious Sherif Abdulatif Wenbin Shi Tobias Hepp Sergios Gatidis Bin Yang Institute of Signal Processing and System Theory University of Stuttgart Stuttgart Germany University of Tübingen Tübingen Germany Max Planck Institute for Intelligent Systems Tübingen Germany

ISBN: (纸本)9781728176055;9781728176062

Age is an essential factor in modern diagnostic procedures. However, assessment of the true biological age (BA) remains a daunting task due to the lack of reference ground-truth labels. Current BA estimation approaches are either restricted to skeletal images or rely on non-imaging modalities that yield a whole-body BA assessment. However, various organ systems may exhibit different aging characteristics due to lifestyle and genetic factors. In this initial study, we propose a new framework for organ-specific BA estimation utilizing 3D magnetic resonance image (MRI) scans. As a first step, this framework predicts the chronological age (CA) together with the corresponding patient-dependent aleatoric uncertainty. An iterative training algorithm is then utilized to segregate atypical aging patients from the given population based on the predicted uncertainty scores. In this manner, we hypothesize that training a new model on the remaining population should approximate the true BA behavior. We apply the proposed methodology on a brain MRI dataset containing healthy individuals as well as Alzheimer's patients. We demonstrate the correlation between the predicted BAs and the expected cognitive deterioration in Alzheimer's patients.

关键词： Training Uncertainty Magnetic resonance imaging Sociology Estimation signal processing algorithms Magnetic resonance

来源：评论

学校读者我要写书评

暂无评论

Towards Analytically Computable Quality Classes for MCG Sensor systems

Towards Analytically Computable Quality Classes for MCG Sens...

引用

作者： Engelhardt, Erik Zaman, Adrian Elzenheimer, Eric Frey, Norbert Schmidt, Gerhard Digital Signal Processing and System Theory Institute of Electrical Engineering and Information Technology Faculty of Engineering Kiel University Kaiserstr. 2 Kiel24143 Germany Medical Department Iii Specialised in Cardiology and Angiology University Medical Center Schleswig-Holstein Arnold-Heller-Str. 3/House K3 Kiel24105 Germany Internal Medicine III: Heart Vascular and Lung University Medical Center Heidelberg Im Neuenheimer Feld 410 Heidelberg69120 Germany

While metrics such as SNR (signal-to-noise ratio) and ASC (application-specific capacity) can be used to describe signal quality quantitatively, they do not allow qualitative assessment. If only the SNR of a signal is given, it is not clear for which applications it is suitable. The subjective evaluation of signals by application experts, on the other hand, can provide such a qualitative assessment, but it is not reproducible. In this study, we investigate the relationship between these easily computable metrics such as SNR and ASC and the subjective evaluation of cardiologists using MCG signals as an example. To do this, we define four quality classes and generate a collection of noisy prototype signals. The cardiologists' assessments are then compared with the SNR and ASC of each signal. Due to the small amount of data, we only use a linear regression to predict quality classes based on common quantitative metrics. The achieved coefficients of determination are RSNR 2 = 0.612 and RASC 2 = 0.593. In perspective, it is therefore plausible to analytically calculate the quality class of a MCG sensor system from the system characteristics with the support of a larger cohort of cardiologists. This will allow sensor system manufacturers to optimize their systems more efficiently for specific applications, enabling faster innovation loops at lower cost. For users of sensor systems, it will simplify the selection of the most suitable system for their application. © 2022 The Author(s), published by De Gruyter.

关键词： signal to noise ratio

来源：评论

学校读者我要写书评

暂无评论

A Deep Learning Framework for Improving Lameness Identification in Dairy Cattle 35

A Deep Learning Framework for Improving Lameness Identificat...

引用

35th AAAI Conference on Artificial Intelligence, AAAI 2021

作者： Karoui, Yasmine Boatswain Jacques, Amanda A. Diallo, Abdoulaye Baniré Shepley, Elise Vasseur, Elsa Laboratoire de Combinatoire Informatique Mathématique Université du Québec à Montréal PO BOX 8888 Downtown Station H3C 3P8 Canada Institute for Signal Processing and System Theory University of Stuttgart Stuttgart Germany McGill University Department of Animal Science Québec Canada

ISBN: (纸本)9781713835974

Lameness, characterized by an anomalous gait in cows due to a dysfunction in their locomotive system, is a serious welfare issue for cows and farmers. Prompt lameness detection methods can prevent the development of acute lameness in cattle. In this study, we propose a deep learning framework to help identify lameness based on motion curves of different leg joints on the cow. The framework combines data augmentation and a convolutional neural network using an LeNet architecture. Performance assessed using cross validation showed promising prediction accuracies above 99% and 91% for validation and test sets, respectively. This also demonstrates the usefulness of data generation in cases where the data set is originally small in size and difficult to generate. Copyright © 2021, Association for the Advancement of Artificial Intelligence (***). All rights reserved

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Uncertainty-based biological age estimation of brain MRI scans

arXiv

引用

arXiv 2021年

作者： Armanious, Karim Abdulatif, Sherif Shi, Wenbin Hepp, Tobias Gatidis, Sergios Yang, Bin University of Stuttgart Institute of Signal Processing and System Theory Stuttgart Germany University of Tübingen Department of Radiology Tübingen Germany Max Planck Institute for Intelligent Systems Empirical Inference Department Tübingen Germany

Age is an essential factor in modern diagnostic procedures. However, assessment of the true biological age (BA) remains a daunting task due to the lack of reference ground-truth labels. Current BA estimation approaches are either restricted to skeletal images or rely on non-imaging modalities that yield a whole-body BA assessment. However, various organ systems may exhibit different aging characteristics due to lifestyle and genetic factors. In this initial study, we propose a new framework for organ-specific BA estimation utilizing 3D magnetic resonance image (MRI) scans. As a first step, this framework predicts the chronological age (CA) together with the corresponding patient-dependent aleatoric uncertainty. An iterative training algorithm is then utilized to segregate atypical aging patients from the given population based on the predicted uncertainty scores. In this manner, we hypothesize that training a new model on the remaining population should approximate the true BA behavior. We apply the proposed methodology on a brain MRI dataset containing healthy individuals as well as Alzheimer’s patients. We demonstrate the correlation between the predicted BAs and the expected cognitive deterioration in Alzheimer’s patients. Copyright © 2021, The Authors. All rights reserved.

关键词： Magnetic resonance imaging

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：