检索结果-内蒙古大学图书馆

arXiv 2020年

作者： Vogt, Ryan Touzel, Maximilian Puelma Shlizerman, Eli Lajoie, Guillaume Department of Applied Mathematics University of Washington Math & Stats Dept. Université de Montréal Mila Québec AI Institute Depts. of App. Math. and Elec. & Comp. Engineering University of Washington

Recurrent neural networks (RNNs) have been successfully applied to a variety of problems involving sequential data, but their optimization is sensitive to parameter initialization, architecture, and optimizer hyperparameters. Considering RNNs as dynamical systems, a natural way to capture stability, i.e., the growth and decay over long iterates, are the Lyapunov Exponents (LEs), which form the Lyapunov spectrum. The LEs have a bearing on stability of RNN training dynamics because forward propagation of information is related to the backward propagation of error gradients. LEs measure the asymptotic rates of expansion and contraction of nonlinear system trajectories, and generalize stability analysis to the time-varying attractors structuring the non-autonomous dynamics of data-driven RNNs. As a tool to understand and exploit stability of training dynamics, the Lyapunov spectrum fills an existing gap between prescriptive mathematical approaches of limited scope and computationally-expensive empirical approaches. To leverage this tool, we implement an efficient way to compute LEs for RNNs during training, discuss the aspects specific to standard RNN architectures driven by typical sequential datasets, and show that the Lyapunov spectrum can serve as a robust readout of training stability across hyperparameters. With this exposition-oriented contribution, we hope to draw attention to this under-studied, but theoretically grounded tool for understanding training stability in RNNs. Copyright © 2020, The Authors. All rights reserved.

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

Optical Lattice with Torus Topology

arXiv

引用

arXiv 2018年

作者： Kim, Hwanmun Zhu, Guanyu Porto, J.V. Hafezi, Mohammad Joint Quantum Institute NIST/University of Maryland College ParkMD20742 United States Department of Physics University of Maryland College ParkMD20742 United States Depts. of Elec. and Comp. Engineering and Institute for Research in Electronics and Applied Physics University of Maryland College ParkMD20742 United States

We propose an experimental scheme to construct an optical lattice where the atoms are confined to the surface of a torus. This construction can be realized with spatially shaped laser beams which could be realized with recently developed high resolution imaging techniques. We numerically study the feasibility of this proposal by calculating the tunneling strengths for atoms in the torus lattice. To illustrate the non-trivial role of topology in atomic dynamics on the torus, we study the quantized superfluid currents and fractional quantum Hall (FQH) states on such a structure. For FQH states, we numerically investigate the robustness of the topological degeneracy and propose an experimental way to detect such a degeneracy. Our scheme for torus construction can be generalized to surfaces with higher genus for exploration of richer topological physics. Copyright © 2018, The Authors. All rights reserved.

关键词： Atoms

来源：评论

学校读者我要写书评

暂无评论

Which parts of the face give out your identity?

Which parts of the face give out your identity?

引用

作者： Ocegueda, Omar Shah, Shishir K. Kakadiaris, Ioannis A. Computational Biomedicine Lab Depts. of Computer ScienceElec. and Comp.Engineering and Biomedical Engineering Univ. of Houston Houston TX United States

ISBN: (纸本)9781457703942

We present a Markov Random Field model for the analysis of lattices (e.g., images or 3D meshes) in terms of the discriminative information of their vertices. The proposed method provides a measure field that estimates the probability of each vertex to be discriminative or non-discriminative. As an application of the proposed framework, we present a method for the selec.ion of compact and robust features for 3D face recognition. The resulting signature consists of 360 coefficients, based on which we are able to build a classifier yielding better recognition rates than currently reported in the literature. The main contribution of this work lies in the development of a novel framework for feature selec.ion in scenarios in which the most discriminative information is known to be concentrated along piece-wise smooth regions of a lattice. © 2011 IEEE.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

ExoGrip: Assisted hand strength glove

ExoGrip: Assisted hand strength glove

引用

作者： Best, Jade E. Bostick, Nehemiah F. Connelly, John R. Dunn, Michael G. Gelles, Richard A. Norvell, Elizabeth K. Waugaman, William B. Mims, Willie H. Depts. of Elec. and Computer Engineering Mechanical Engineering Systems Engineering Management U. S. Air Force Academy CO 80840 United States

ISBN: (纸本)9781617827396

A large number of American troops fighting in Afghanistan and Iraq have received wounds in their upper extremities leading to significant nerve damage and loss of strength. These injuries impair their ability to perform day-to-day tasks such as lifting a cup of coffee or opening a door. Although the cause of some injuries in service-people is often unique to their employment, civilian employees in other industries are also plagued with similar physical damage due to other kinds of injuries. Our goal is to develop a device to augment the strength of injured troops and civilian workers so they can perform everyday tasks despite their physical limitations. The ExoGrip is a glove designed to provide this necessary strength augmentation. The ExoGrip consists primarily of pressure sensors, linear actuators, and a microcontroller to provide a force multiplier based on a person's strength. The goal of the first phase of the project was to conduct research and also produce a working prototype of one finger. This goal was achieved by a group of classmates who started the project a year before. Their research and feasibility analysis ended in the mechanical movement of a single finger when the sensors were activated. The next phase of this project is to design and integrate a working prototype that manipulates all four fingers, while keeping the thumb in a fixed position. This paper describes the integration of new microcontrollers, linear actuators utilizing pulse width modulation technology, and improved pressure sensors needed to manipulate the fingers, as well as laying the foundation for future testing and development of a final product.

关键词： Microcontrollers

来源：评论

学校读者我要写书评

暂无评论

A shape-driven MRF model for the segmentation of organs in medical images

A shape-driven MRF model for the segmentation of organs in m...

引用

Conference on computer Vision and Pattern Recognition (CVPR)

作者： D. R. Chittajallu S. K. Shah I. A. Kakadiaris Computational Biomedicine Lab Depts. of Computer Science Elec. & Comp.Engineering and Biomedical Engineering University of Houston Houston TX USA

In this paper, we present a knowledge-driven Markov Random Field (MRF) model for the segmentation of organs in medical images with particular emphasis on the incorporation of shape constraints into the segmentation problem. We cast the problem of image segmentation as the Maximum A Posteriori (MAP) estimation of a Markov Random Field which, in essence, is equivalent to the minimization of the corresponding Gibbs energy function. We then incorporate a set of constraints into the Gibbs energy function that collectively force the resulting segmentation contour/surface to have a shape similar to that of a given shape template. In particular, we introduce a flux-maximization constraint and a generalized template-based star-shape constraint that are encoded into the first- and second-order clique potentials of the Gibbs energy function, respectively. Our main contribution is in the translation of a set of global notions about the shape of the desired segmentation contour into a set of local measures that can be conveniently encoded into the Gibbs energy function and used in combination with other traditionally used constraints derived from image information. In our experiments, we demonstrate the application of the proposed method to the challenging problem of heart segmentation in non-contrast computed tomography (CT) data.

关键词： Image segmentation Biomedical imaging Markov random fields Shape measurement Computed tomography Biomedical engineering Energy measurement Heart Biomedical computing computer science

来源：评论

学校读者我要写书评

暂无评论

Predicting cardiovascular disease from real-time elec.rocardiographic monitoring: An adaptive machine learning approach on a cell phone

Predicting cardiovascular disease from real-time electrocard...

引用

31st Annual International Conference of the IEEE engineering in Medicine and Biology Society: engineering the Future of Biomedicine, EMBC 2009

作者： Jin, Zhanpeng Sun, Yuwen Cheng, Allen C. Department of Electrical and Computer Engineering University of Pittsburgh Pittsburgh PA 15261 United States Depts. of Elec. and Computer Engineering Bioengineering Neurological Surgery and Comp. Science University of Pittsburgh Pittsburgh PA 15261 United States

ISBN: (纸本)9781424432967

To date, cardiovascular disease (CVD) is the leading cause of global death. The elec.rocardiogram (ECG) is the most widely adopted clinical tool that measures the elec.rical activities of the heart from the body surface. However, heart rhythm irregularities cannot always be detected on a standard resting ECG machine, since they may not occur during an individual's recording session. Common Holter-based portable solutions that record ECG for up to 24 to 48 hours lack the capability to provide real-time feedback. In this research, we seek to establish a cell phone-based real-time monitoring technology for CVD, capable of performing continuous on-line ECG processing, generating a personalized cardiac health summary report in layman's language, automatically detecting and classifying abnormal CVD conditions, all in real time. Specifically, we developed an adaptive artificial neural network (ANN)-based machine learning technique, combining both an individual's cardiac characteristics and information from clinical ECG databases, to train the cell phone to learn to adapt to its user's physiological conditions to achieve better ECG feature extraction and more accurate CVD classification on cell phones. ©2009 IEEE.

关键词： elec.rocardiograms

来源：评论

学校读者我要写书评

暂无评论

Current Measurement Gain Tuning Using High Frequency Signal Injection

Current Measurement Gain Tuning Using High Frequency Signal ...

引用

Conference Record of the IEEE Industry Applications Society Annual Meeting (IAS)

作者： Michael C. Harke Juan M. Guerrero Michael W. Degner Fernando Briz Robert D. Lorenz Hamilton Sundstrand Applied Research Rockford IL USA Dept. of Electrical Computer & Systems Engineering Gijón Spain Ford Motor Company Research and Advanced Engineering Dearborn MI USA Department of Electrical Computer & Systems Engineering University of Oviedo Gijon Spain Depts. of Mech. Engr. and Elec. & Compo. Engr. University of Wisconsin - Madison Madison WI USA

Accurate measurement of the phase currents in elec.ric machines is a requirement for most high performance drives. Incorrect current measurement gain estimates, caused by device tolerances, temperature drifts, age and noise, reduce the accuracy of the current measurements and degrade the drive's performance. This paper presents a method for compensating current measurement gain errors using a periodically injected, high frequency voltage. The positive and negative sequence components of the resulting high frequency current are measured and used for balancing and scaling the phase current measurement gains. The method can be used either as part of an initial commissioning, or start-up sequence, or periodically during the regular operation of the machine.

关键词： Current measurement Tuning Frequency Phase measurement elec.ric variables measurement elec.ric machines Performance gain Temperature Noise reduction Degradation

来源：评论

学校读者我要写书评

暂无评论

Saliency tracking-based, sensorless control of AC machines using structured neural networks

Saliency tracking-based, sensorless control of AC machines u...

引用

2005 IEEE Industry Applications Conference, 40th IAS Annual Meeting

作者： García, Pablo Briz, Fernando Racat, Dejan Lorenz, Robert D. University of Oviedo Dept. of Elec. Computer and Systems Engineering Gijón 33204 Spain Depts. of Mech. Engr. and Elect. and Comp. Engr. 1415 Engineering Drive Madison WI 53706

ISBN: (纸本)0780392086

The focus of this paper is the use of structured neural networks for sensorless control of AC machines using the zero sequence carrier signal voltage. Structured neural networks allow effective compensation of saturation-induced saliencies as well as other secondary saliencies. In comparison with classical compensation methods, such as lookup tables, this technique has advantages such as physics-based structure (and thus is potentially insightful), general scalability, reduced size and complexity, and correspondingly reduced commissioning time. When compared with traditional neural network solutions, the structured neural networks are simpler, physically insightful, less computationally intensive and easier to train. All make the proposed method an improved implementation for sensorless drives. © 2005 IEEE.

关键词： AC machinery

来源：评论

学校读者我要写书评

暂无评论

Blue swarm 3: Integrating capabilities for an autonomous rescue robot swarm

Blue swarm 3: Integrating capabilities for an autonomous res...

引用

19th National Conference on Artificial Intelligence

作者： Stormont, Daniel P. Arora, Anisha Chandrasekharan, Madhumita Datar, Manasi Dave, Udit Gharpure, Chaitanya Jorge, Juan Kutiyanawala, Aliasgar Patel, Vinay Ramaswamy, Bharath Sackett, Loren Song, Zhen Depts. of Comp. Sci. Elec. and Computer Engineering and Mechanical and Aerospace Engineering Utah State University Logan UT 84322

The robotics team at Utah State University has been working on incrementally building the capabilities needed to field an autonomous swarm of rescue robots. We believe a swarm of small, low-cost robots that can identify areas needing further investigation would be of great benefit to first responders to the scene of a disaster. Blue Swarm 3 is the latest iteration in this development process and is the first swarm to try to incorporate all of the capabilities needed for a fieldable rescue robot swarm. This paper discusses the philosophy behind Blue Swarm 3, the historical background behind the previous swarms, details of the design of Blue Swarm 3, the results of our efforts at this year's American Association for Artificial Intelligence Mobile Robot Competition, lessons learned, and plans for the future development of the Blue Swarm 3 and follow-on swarms. Copyright © 2004, American Association for Artificial Intelligence. All rights reserved.

关键词： Robotics

来源：评论

学校读者我要写书评

暂无评论

Improved diagnosis and monitoring of cancer using portable microfluidics platforms

Improved diagnosis and monitoring of cancer using portable m...

引用

Proceedings - 2004 International Conference on MEMS, NANO and Smart Systems, ICMENS 2004

作者： Pilarski, Linda M. Adamia, Sophia Pilarski, Patrick M. Prakash, Ranjit Lauzon, Jana Backhouse, Christopher J. Depts. Oncol. and Elec. Engineering University of Alberta Cross Cancer Institute Alta. Canada

ISBN: (纸本)0769521894

Microfluidics chips offer the promise of fast and accurate diagnostic and monitoring tools that are inexpensive and automated. Tests that otherwise require expensive instrumentation and consume valuable reagents can be implemented on microfluidic chips for a fraction of the cost with dramatic improvement in speed. Minaturization offers multiple benefits, enabling multiple tests on one chip, requiring tiny tissue samples and fast reaction times for picoliter volumes. More complex microfluidics chips will be able to carry out sample processing followed by cell or molecule fractionation, molecular analysis and readout of the test result. Integrated platforms incorporating microfluidics offer portable testing devices for "point of concern" testing in health care facilities, in remote areas or in the home. These devices are likely to increase access to sophisticated testing within the health system while decreasing costs and improving the quality of care. © 2004 IEEE.

关键词： Microelec.ronics

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：