检索结果-内蒙古大学图书馆

A New Approach to Estimate RED Parameters Using function regression

INTERNATIONAL JOURNAL OF FUTURE GENERATION COMMUNICATION AND NETWORKING 2014年第4期7卷 103-118页

作者： Wang, Wentao Wang, Hao Tang, Wan Guo, Feng South Cent Univ Nationalities Coll Comp Sci 182 Minvuan Rd Wuhan 430074 Hubei Peoples R China

Random Early Drop (RED) is widely applied in Mobile Ad-hoc NETworks (MANETs) for congestion control. It randomly drops packets to prevent congestion from occurring, while keeping proper queue size of Route Request (RREQ) packets at the same time. Unfortunately, in the case of severe congestion and large amounts of RREQ packets in the queue, RED algorithm with tune parameters is not able to improve the congestion status properly. This paper proposes a Dynamic RED (DRED) mechanism based on the fitting curve of packet delivery ratio and packet sending rate, which can be readily achieved in practice. DRED presupposes threshold value for initiating and dynamically changes it according to the function regression. Compared to First In First Out (FIFO) and RED with tune parameters mechanisms, simulation results show that DRED has better performance in terms of average end-to-end delay, Hello packet overhead and packet delivery ratio, while avoiding obvious increase of routing discovery frequency.

关键词： Ad-hoc NETworks congestion control function regression RED

来源：评论

学校读者我要写书评

暂无评论

ANALYSIS OF A TWO-LAYER NEURAL NETWORK VIA DISPLACEMENT CONVEXITY

引用

ANNALS OF STATISTICS 2020年第6期48卷 3619-3642页

作者： Javanmard, Adel Mondelli, Marco Montanari, Andrea Univ Southern Calif Marshall Sch Business Data Sci & Operat Dept Los Angeles CA 90007 USA IST Austria Klosterneuburg Austria Stanford Univ Dept Elect Engn Stanford CA 94305 USA Stanford Univ Dept Stat Stanford CA 94305 USA

Fitting a function by using linear combinations of a large number N of "simple" components is one of the most fruitful ideas in statistical learning. This idea lies at the core of a variety of methods, from two-layer neural networks to kernel regression, to boosting. In general, the resulting risk minimization problem is nonconvex and is solved by gradient descent or its variants. Unfortunately, little is known about global convergence properties of these approaches. Here, we consider the problem of learning a concave function f on a compact convex domain Omega subset of R-d, using linear combinations of "bump-like" components (neurons). The parameters to be fitted are the centers of N bumps, and the resulting empirical risk minimization problem is highly nonconvex. We prove that, in the limit in which the number of neurons diverges, the evolution of gradient descent converges to a Wasserstein gradient flow in the space of probability distributions over Omega. Further, when the bump width delta tends to 0, this gradient flow has a limit which is a viscous porous medium equation. Remarkably, the cost function optimized by this gradient flow exhibits a special property known as displacement convexity, which implies exponential convergence rates for N -> infinity, delta -> 0. Surprisingly, this asymptotic theory appears to capture well the behavior for moderate values of delta, N. Explaining this phenomenon, and understanding the dependence on delta, N in a quantitative manner remains an outstanding challenge.

关键词： Neural networks stochastic gradient descent Wasserstein gradient flow function regression convergence rate displacement convexity

来源：评论

学校读者我要写书评

暂无评论

Linking species richness and size diversity in birds and fishes

引用

ECOGRAPHY 2018年第12期41卷 1979-1991页

作者： Yen, Jian D. L. Thomson, James R. Keith, Jonathan M. Paganin, David M. Fleishman, Erica Bennett, Andrew F. Nimmo, Dale G. Bennett, Joanne M. Dobkin, David S. Mac Nally, Ralph Univ Melbourne Sch Biosci Melbourne Vic 3010 Australia Univ Melbourne ARC Ctr Excellence Environm Decis Melbourne Vic Australia Arthur Rylah Inst Environm Res Dept Environm Land Water & Planning Heidelberg Vic Australia La Trobe Univ Dept Ecol Environm & Evolut Bundoora Vic Australia Monash Univ Sch Math Sci Melbourne Vic Australia Monash Univ Sch Phys & Astron Melbourne Vic Australia Univ Calif Davis John Muir Inst Environm Davis CA 95616 USA Colorado State Univ Dept Fish Wildlife & Conservat Biol Ft Collins CO 80523 USA La Trobe Univ Dept Ecol Environm & Evolut Bundoora Vic Australia Univ Canberra Inst Appl Ecol Canberra ACT Australia Charles Sturt Univ Sch Environm Sci Inst Land Water & Soc Albury NSW Australia German Ctr Integrat Biodivers Res iDiv Leipzig Germany Martin Luther Univ Halle Wittenberg Inst Biol Halle Saale Germany High Desert Ecol Res Inst Bend OR USA

The rapid development of mechanistic, trait-based models has resulted in increasingly reliable predictions of the functional diversity of individuals in populations and communities. However, a focus on individuals' traits differs from the prevailing focus on species in much of community ecology. We sought to identify correlative links between species richness and size diversity, focusing on size diversity as one component of functional diversity. These links could be used to extend individual, size-based models to predict patterns of species richness. We used the distribution of the sizes of individuals in a community - the individual-size distribution (ISD) - as a measure of size diversity, and constructed Bayesian regression models with species richness as the response variable and ISDs as the predictor variables. We used two methods to include ISDs in our analyses. First, we summarized the ISD with five common diversity indices and used these indices as predictor variables in our analyses. Second, we used functional data analysis to include the entire ISD (a continuous function) as a predictor variable in our analyses. Analyses of diversity indices identified consistent, positive associations between species richness and size diversity. Analyses of entire ISDs revealed that these associations were driven by numbers of small- and medium-sized individuals. In general, a combination of diversity indices predicted species richness as well as or better than continuous ISDs. However, models with ISDs as predictor variables were less sensitive to technical details of model fitting (e.g. discretization method) than those based on diversity indices, and the use of ISDs avoids the arbitrary selection of one or several diversity indices. Our use of functional data analysis allows any trait distribution to be included as a variable in statistical analyses, and has the potential to reveal new diversity patterns in ecology.

关键词： biological diversity function regression size spectrum

来源：评论

学校读者我要写书评

暂无评论

A biogeographical perspective on species abundance distributions: recent advances and opportunities for future research

引用

JOURNAL OF BIOGEOGRAPHY 2017年第8期44卷 1705-1710页

作者： Matthews, Thomas J. Borges, Paulo A. V. de Azevedo, Eduardo Brito Whittaker, Robert J. Univ Birmingham GEES Sch Geog Earth & Environm Sci Birmingham B15 2TT W Midlands England CE3C Azorean Biodivers Grp P-9700042 Angra Do Heroismo Acores Portugal Univ Acores Dept Ciencias & Engn Ambiente P-9700042 Angra Do Heroismo Acores Portugal Univ Acores Dept Ciencias Agr CMMG Ctr Estudos Clima Meteorol & Mudancas Globai CITA A P-9700042 Angra Do Heroismo Portugal Univ Oxford Sch Geog & Environm Conservat Biogeog & Macroecol Programme South Parks Rd Oxford OX1 3QY England Univ Copenhagen Dept Biol Ctr Macroecol Evolut & Climate Univ Pk 15 DK-2100 Copenhagen O Denmark

It has become increasingly recognized that multiple processes can generate similar shapes of species abundance distributions (SADs), with the result that the fit of a given SAD model cannot unambiguously provide evidence in support of a given theory or model. An alternative approach to comparing the fit of different SAD models to data from a single site is to collect abundance data from a variety of sites, and then build models to analyse how different SAD properties (e.g. form, skewness) vary with different predictor variables. Such a biogeographical approach to SAD research is potentially very revealing, yet there has been a general lack of interest in SADs in the biogeographical literature. In this Perspective, we address this issue by highlighting findings of recent analyses of SADs that we consider to be of intrinsic biogeographical interest. We use arthropod data drawn from the Azorean archipelago to further highlight how analyses of SAD form and function may be biogeographically informative. We hope that, by reviewing a number of novel approaches, our article may prove to be a catalyst for a greater interest in analysing SADs in biogeography.

关键词： applied biogeography arthropods Azores function regression gambin model lognormal model logseries model species abundance distribution

来源：评论

学校读者我要写书评

暂无评论

Use ANN to Solve function regression Problem in Manufacturing

引用

International Conference on Frontiers of Manufacturing and Design Science (ICFMD2010)

作者： Fan, Qi Zhu, Chang-jie Wang, Bao-hua Xiao, Jian-yu Huaibei Normal Univ Sch Comp Sci & Technol Huaibei City 235000 Anhui Peoples R China

ISBN: (纸本)9783037850046

Engineers often meet function regression problems in manufacturing. In this work, we use artifical neural network to solve this problem. We choose a typical function as the target function. Since the input value of the function is continuous, the output of the regression model should also have a continuous value range. We implement the feed forward neural network with back propagation learning algorithm, investigate different network parameters, and compare several different training algorithms. The performance assessment based on the test dataset is also discussed.

关键词： Artifical Neural network function regression Feed Forward Neural network Back propagation

来源：评论

学校读者我要写书评

暂无评论

Study on SVM on-line function regression method for mass data

Study on SVM on-line function regression method for mass dat...

引用

6th International Conference on Machine Learning and Cybernetics

作者： An, Jin-Long Yang, Qing-Xin Ma, Zhen-Ping Hebei Univ Technol Prov Minist Joint Key Lab Electromagnet Field & E Tianjin 300130 Peoples R China

ISBN: (纸本)9781424409723

In order to overcome the problems that the SVM training time is too long for a large number of samples and that SVM cannot be trained online when the samples increase dynamically, a new approach of SVM online function regression for mass samples is put forward in this paper. And the validity of this method is proved by simulation experiment.

关键词： online support vector machine function regression

来源：评论

学校读者我要写书评

暂无评论

STUDY ON SVM ON-LINE function regression METHOD FOR MASS DATA

STUDY ON SVM ON-LINE FUNCTION REGRESSION METHOD FOR MASS DAT...

引用

2007 International Conference on Machine Learning and Cybernetics(IEEE第六届机器学习与控制论国际会议)

作者： JIN-LONG AN QING-XIN YANG ZHEN-PING MA Province-Ministry Joint Key Laboratory Of Electromagnetic Field And Electrical Apparatus Reliability Hebei University Of Technology Tianjin China300130

In order to overcome the problems that the SVM training time is too long for a large number of samples and that SVM cannot be trained online when the samples increase dynamically, a new approach of SVM online function regression for mass samples is put forward in this *** the validity of this method is proved by simulation experiment.

关键词： Online Support vector machine function regression

来源：评论

学校读者我要写书评

暂无评论

Support Vector Machine Filtering for MEMS gyroscope

Support Vector Machine Filtering for MEMS gyroscope

引用

第三届国际自动化测试与仪器仪表学术会议

作者： Yang Peng,Li Qing (Key Laboratory of Modern Measurement &Control Technology,University of Beijing Information Science & Technology,Beijing 100192 China)

Least squares support vector machine is new methods of classification and regression function in the field of machine learning in recent *** paper introduces the basic principles and algorithm of least squares support vector *** realize the least squares support vector machines' application in real-time filtering prediction for the gyro's random *** the MATLAB simulation,the results show this method is effective.

关键词： least squares support vector machine function regression MEMS gyroscope real-time filtering

来源：评论

学校读者我要写书评

暂无评论

Auto-correlation wavelet support vector machine

引用

IMAGE AND VISION COMPUTING 2009年第8期27卷 1040-1046页

作者： Chen, G. Y. Dudek, G. McGill Univ Ctr Intelligence Machines Montreal PQ H3A 2A7 Canada

A support vector machine (SVM) with the auto-correlation of a compactly supported wavelet as a kernel is proposed in this paper. The authors prove that this kernel is an admissible support vector kernel. The main advantage of the auto-correlation of a compactly supported wavelet is that it satisfies the translation invariance property, which is very important for its use in signal processing. Also, we can choose a better wavelet by selecting from different wavelet families for our auto-correlation wavelet kernel. This is because for different applications we should choose wavelet filters selectively for the autocorrelation kernel. We should not always select the same wavelet fllters independent of the application, as we demonstrate. Experiments on signal regression and pattern recognition show that this kernel is a feasible kernel for practical applications. (C) 2008 Elsevier B.V. All rights reserved.

关键词： Wavelets Support vector machine Machine learning Pattern recognition function regression Auto-correlation

来源：评论

学校读者我要写书评

暂无评论

Parameter optimization of LS-SVM for regression using NGA

Parameter optimization of LS-SVM for regression using NGA

引用

Conference on Intelligent Computing - Theory and Applications V

作者： Wang, Qi Feng, Zhigang SHIDA, Katsunori Harbin Inst Technol Dept Automat Measurement & Control Harbin Peoples R China

ISBN: (纸本)9780819466822

Compared with support vector machine (SVM), least squares support vector machine (LS-SVM) has overcame the shortcoming of higher computational burden by solving linear equations, and has been widely used in classification and nonlinear function estimation. But there is no efficient method for parameter selection of LS-SVM. In this paper, the sharing function based niche genetic algorithm (SNGA) is used to the parameter optimization of LS-SVM for regression. In the SNGA approach, k-folds cross validation is used to evaluate the LS-SVM generalization performance. The inverse of the average test error of the k trials is used as the fitness value. The hamming distance between each two individuals is defined as the sharing function. Two benchmark problems, SINC function regression and Henon map time series prediction are used as examples for demonstration. The results indicate that this approach can escape from the blindness of man-made choice of the LS-SVM parameters. It enhances the efficiency and the capability of regression. With little modification, this approach is also can be used to the parameter optimization of SVM or LS-SVM for classification.

关键词： parameter optimization function regression least squares support vector machine niche genetic algorithm k-folds cross validation

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：