检索结果-内蒙古大学图书馆

9th IEEE International Conference on Intelligent Computing and Information Systems, ICICIS 2019

作者： Aly, Ahmed Guadagni, Gianluca Dugan, Joanne B. Computer Engineering Program University of Virginia CharlottesvilleVA United States University of Virginia Applied Math Department CharlottesvilleVA United States

ISBN: (纸本)9781728139951

Deep neural networks (DNNs) have been found useful for many applications. However, training and designing those networks can be challenging and is considered more of an art or an engineering process than rigorous science. In this regard, the important process of choosing hyperparameters is relevant. In addition, training neural networks with derivative-free methods is somewhat understudied. Particularly, with regards to hyperparameter selection. The paper presents a small-scale study of 3 hyperparameters choice for convolutional neural networks (CNNs). The networks were trained with two single-candidate optimization algorithms: Stochastic Gradient Descent (derivative-based) and Local Search (derivative-free). The CNN is trained on a subset of the FashionMNIST dataset. Experimental results show that hyperparameter selection can be detrimental for Local Search, especially regarding network parametrization. Moreover, the best hyperparameter choices didn't match for both algorithms. Future investigation into the training dynamics of Local Search is likely needed. © 2019 IEEE.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

New Amplitude Equations for Thin Elastic Rods

引用

Physical Review Letters 1996年第17期77卷 3537-3537页

作者： Alain Goriely Michael Tabor Program in Applied Mathematics University of Arizona Math Building #89 Tucson Arizona 85721

The stability of twisted straight rods is described within the framework of the time dependent Kirchhoff equations for thin elastic filaments. A perturbation method is developed to study the linear stability of this problem and find the dispersion relations. A nonlinear analysis results in a new amplitude equation, describing the deformation of the rod beyond the instability, which takes the form of a pair of nonlinear, second-order evolution equations coupling the local deformation amplitude to the twist density. Various solutions, such as solitary waves, are presented.

关键词： .Nonlinear Analysis laments thin elastic New Amplitude twisted straight

来源：评论

学校读者我要写书评

暂无评论

Simple Solution to the Nonlinear Front Problem

引用

Physical Review Letters 1995年第10期75卷 2047-2047页

作者： Alain Goriely Program in Applied Mathematics University of Arizona Math Building No. 89 Tucson Arizona 85721

The problem of determining the front speed for one-dimensional real reaction-diffusion equations is considered. A new solution to the problem, valid for a large class of functions, is proposed. In contrast with other methods, this new approach does not rely on the explicit computation of the front solutions and provides an explicit formula relating the nonlinear speed to the parameters of the equation.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Derivative-Free Optimization of Neural Networks using Local Search

Derivative-Free Optimization of Neural Networks using Local ...

引用

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

作者： Ahmed Aly Gianluca Guadagni Joanne B. Dugan Computer Engineering Program University of Virginia Charlottesville VA Applied Math Program University of Virginia Charlottesville VA

ISBN: (纸本)9781728138862

Deep Neural Networks have received a great deal of attention in the past few years. Applications of Deep Learning broached areas of different domains such as Reinforcement Learning and Computer Vision. Despite their popularity and success, training neural networks can be a challenging process. This paper presents a study on derivative-free, single-candidate optimization of neural networks using Local Search (LS). LS is an algorithm where constrained noise is iteratively applied to subsets of the search space. It is coupled with a Score Decay mechanism to enhance performance. LS is a subsidiary of the Random Search family. Experiments were conducted using a setup that is both suitable for an introduction of the algorithm and representative of modern deep learning tasks, based on the FashionMNIST dataset. Training of a 5-Million parameter CNN was done in several scenarios, including Stochastic Gradient Descent (SGD) coupled with Backpropagation (BP) for comparison. Results reveal that although LS was not competitive in terms of convergence speed, it was actually able to converge to a lower loss than SGD. In addition, LS trained the CNN using Accuracy rather than Loss as a learning signal, though to a lower performance. In conclusion, LS presents a viable alternative in cases where SGD fails or is not suitable. The simplicity of LS can make it attractive to non-experts who would want to try neural nets for the first-time or on novel, non-differentiable tasks.

关键词： gradient methods learning (artificial intelligence) neural nets optimisation search problems stochastic processes Neural network tabu search gradient methods Markov chain Stochastic Processes

来源：评论

学校读者我要写书评

暂无评论

Exact solutions for the Saffman-Taylor problem with surface tension

引用

Physical Review Letters 1990年第24期65卷 2986-2986页

作者： Leo P. Kadanoff Computational and Applied Math Program Ryerson Laboratory The University of Chicago 1100 East 58th Street Chicago Illinois 60637

Solutions are reported for the problem of the motion of a two-dimensional interface between a viscous and a nonviscous fluid. The solution has the interface moving uniformly while the viscous fluid has a nontrivial potential flow. In an alternative interpretation, there are gravitational forces in the plane which allow the interface to be at rest while the fluid is in motion.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Initial Energy of a Spatially Flat Universe - a Hint of its Possible Origin

arXiv

引用

arXiv 2022年

作者： Melia, Fulvio Department of Physics and Astronomy The Applied Math Program The University of Arizona TucsonAZ United States

The evidence for a Big Bang origin of the Universe is truly compelling, though its cause remains a complete mystery. As the cosmic spacetime is revealed to us with ever improving detail, however, we are beginning to refine the range of its possible initial conditions-at least within the framework of current physical theories. The Universe, it seems, is spatially flat, and here we discuss in clear, straightforward terms why this trait implies a cosmos with zero kinetic plus gravitational energy, though apparently not zero total energy. Such an outcome has far reaching consequences because of the possibility that the Universe may have begun its existence as a quantum fluctuation. Was this from 'nothing,' or perhaps a pre-existing vacuum? A non-zero total energy would seemingly preclude the former scenario, but not necessarily the latter, though this would then raise the question of how a fluctuation with non-zero energy could have lived long enough, or classicalized, for us to see it 13.5 billion years later. The high-precision measurement of the Universe's spatial curvature may thus constitute the first tangible piece of evidence impacting a possible quantum beginning. © 2022, CC BY.

关键词： Gravitation

来源：评论

学校读者我要写书评

暂无评论

A resolution of the trans-planckian problem in the Rh=ct universe

arXiv

引用

arXiv 2020年

作者： Melia, Fulvio Departments of Physics and Astronomy The Applied Math Program The University of Arizona TucsonAZ United States

The recent measurement of a cutoff kmin in the fluctuation power spectrum p(k) of the cosmic microwave background may vitiate the possibility that slow-roll inflation can simultaneously solve the horizon problem and account for the formation of structure via the growth of quantum fluctuations in the inflaton field. Instead, we show that kmin may be interpreted more successfully in the Rh = ct cosmology, as the first mode exiting from the Planck scale into the semi-classical Universe shortly after the Big Bang. In so doing, we demonstrate that such a scenario completely avoids the well-known trans-Planckian problem plaguing standard inflationary cosmology. Copyright © 2020, The Authors. All rights reserved.

关键词： Cosmology

来源：评论

学校读者我要写书评

暂无评论

A measurement of the cosmic expansion within our lifetime

arXiv

引用

arXiv 2021年

作者： Melia, Fulvio Department of Physics Applied Math Program The University of Arizona TucsonAZ85721 United States

The most exciting future observation in cosmology will feature a monitoring of the cosmic expansion in real time, unlike anything that has ever been attempted before. This campaign will uncover crucial physical properties of the various constituents in the Universe, and perhaps answer a simpler question concerning whether or not the cosmic expansion is even accelerating. An unambiguous yes/no response to this query will significantly impact cosmology, of course, but also the standard model of particle physics. Here, we discuss-in a straightforward way-how to understand the so-called 'redshift drift' sought by this campaign, and why its measurement will help us refine the standard-model parameters if the answer is 'yes.' A 'no' answer, on the other hand, could be more revolutionary, in the sense that it might provide a resolution of several long-standing problems and inconsistencies in our current cosmological models. An outcome of zero redshift drift, for example, would obviate the need for a cosmological constant and render inflation completely redundant. © 2021, CC BY.

关键词： Expansion

来源：评论

学校读者我要写书评

暂无评论

On the non stationary Navier-Stokes equations with an external force

引用

Advances in Differential Equations 1999年第5期4卷 697-730页

作者： Cannone, Marco Planchon, Fabrice UMR 7599 du CNRS UFR Mathématiques Université Denis Diderot Paris VII 75251 Paris Cedex 05 2 Place Jussieu France Program for Applied and Computational Mathematics Princeton University Princeton NJ 08544-1000 United States

We present two different existence and uniqueness algorithms for constructing global mild solutions in C([0, T);L3(ℝ3)) to the Cauchy problem for the Navier-Stokes equations with an external force.

We present two different existence and uniqueness algorithms for constructing global mild solutions in C([0, T);L³(ℝ³)) to the Cauchy problem for the Navier-Stokes equations with an external force.

关键词： 35Q30 76D03 76D05

来源：评论

学校读者我要写书评

暂无评论

KLF2 maintains lineage fidelity and suppresses CD8 T cell exhaustion during acute LCMV infection

引用

Science (New York, N.Y.) 2025年第6735期387卷 eadn2337页

作者： Fagerberg, Eric Attanasio, John Dien, Christine Singh, Jaiveer Kessler, Emily A. Abdullah, Leena Shen, Jian Hunt, Brian G. Connolly, Kelli A. De Brouwer, Edward He, Jiaming Iyer, Nivedita R. Buck, Jessica Borr, Emily R. Damo, Martina Foster, Gena G. Giles, Josephine R. Huang, Yina H. Tsang, John S. Krishnaswamy, Smita Cui, Weiguo Joshi, Nikhil S. Department of Immunobiology Yale University School of Medicine New Haven CT United States Program in Computational Biology & Bioinformatics Yale University New Haven CT United States Department of Microbiology and Immunology Geisel School of Medicine at Dartmouth NH United States Department of Pathology Feinberg School of Medicine at Northwestern University Chicago IL United States Department of Genetics and Computer Science Yale University School of Medicine New Haven CT United States Department of Medicine University of Chicago Chicago IL United States Section of Hematology Department of Internal Medicine Yale University School of Medicine New Haven CT United States Department of Systems Pharmacology and Translational Therapeutics University of Pennsylvania Philadelphia PA United States Institute for Immunology Perelman School of Medicine University of Pennsylvania Philadelphia PA United States Parker Institute for Cancer Immunotherapy Perelman School of Medicine University of Pennsylvania Philadelphia PA United States Yale Center for Systems and Engineering Immunology Yale University School of Medicine New Haven CT United States New Haven CT United States Department of Biomedical Engineering Yale University New Haven CT United States Computational Biology and Bioinformatics Program Yale University CT USA New Haven United States Applied Math Program Yale University New Haven CT United States

Naïve CD8 T cells have the potential to differentiate into a spectrum of functional states during an immune response. How these developmental decisions are made and what mechanisms exist to suppress differentiation toward alternative fates remains unclear. We employed in vivo CRISPR-Cas9-based perturbation sequencing to assess the role of ~40 transcription factors (TFs) and epigenetic modulators in T cell fate decisions. Unexpectedly, we found that knockout of the TF Klf2 resulted in aberrant differentiation to exhausted-like CD8 T cells during acute infection. KLF2 was required to suppress the exhaustion-promoting TF TOX and to enable the TF TBET to drive effector differentiation. KLF2 was also necessary to maintain a polyfunctional tumor-specific progenitor state. Thus, KLF2 provides effector CD8 T cell lineage fidelity and suppresses the exhaustion program.

关键词：

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：