检索结果-内蒙古大学图书馆

24th European Signal Processing Conference (EUSIPCO)

作者： Combettes, Patrick L. Pesquet, Jean-Christophe UPMC Univ Paris 06 Sorbonne Univ Lab Jacques Louis Lions UMR 7598 Paris France Univ Paris Est Lab Informat Gaspard Monge CNRS UMR 8049 Champs Sur Marne France

ISBN: (纸本)9780992862657

stochastic approximation techniques have been used in various contexts in data science. We propose a stochastic version of the forward-backward algorithm for minimizing the sum of two convex functions, one of which is not necessarily smooth. Our framework can handle stochastic approximations of the gradient of the smooth function and allows for stochastic errors in the evaluation of the proximity operator of the nonsmooth function. The almost sure convergence of the iterates generated by the algorithm to a minimizer is established under relatively mild assumptions. We also propose a stochastic version of a popular primal-dual proximal splitting algorithm, establish its convergence, and apply it to an online image restoration problem.

关键词： convex optimization nonsmooth optimization primal-dual algorithm stochastic algorithm parallel algorithm proximity operator recovery image restoration

来源：评论

学校读者我要写书评

暂无评论

Application of inverse analysis to determine the geometric configuration of filament heaters for uniform heating

引用

INTERNATIONAL JOURNAL OF THERMAL SCIENCES 2016年 105卷 1-12页

作者： Lemos, Larissa D. Brittes, Rogerio Franca, Francis H. R. Univ Fed Rio Grande do Sul Dept Mech Engn Rua Sarmento Leite 425 BR-90050170 Porto Alegre RS Brazil Univ Fed Santa Maria Av Presidente Vargas 1958 BR-96506302 Cachoeira Do Sul RS Brazil

In typical designs of ovens, it is often desired to determine the position and geometrical configurations of heaters to attain uniform heating of the materials subjected to thermal processes. In this study, uniform temperature and heat flux are prescribed in one of the surfaces of a radiative system, while the shape and position of filament heaters are left to be found. This is equivalent to the inverse problem of the conventional trial-and-error approach. The oven is modeled as a three-dimensional enclosure formed with diffuse-gray walls, and filled with transparent medium. The inverse problem is solved as an optimization problem. The solution is obtained by the generalized extreme optimization (GEO) algorithm, a stochastic global optimization method. The proposed methodology can lead to filament heaters capable of satisfying the two conditions on the design surface with maximum deviation of about 2% and average deviation less than 0.5%. (C) 2016 Elsevier Masson SAS. All rights reserved.

关键词： Uniform heating Filament heater Inverse design stochastic algorithm

来源：评论

学校读者我要写书评

暂无评论

Computer Simulator of Glycine Receptor Activity: A New Window into a Virtual World

引用

BULLETIN OF MATHEMATICAL BIOLOGY 2016年第7期78卷 1380-1393页

作者： Boronovsky, S. E. Nartsissov, Y. R. Inst Cytochem & Mol Pharmacol Dept Math Modelling & Stat Anal Moscow 115404 Russia

Our study reports computer software that simulates the work of a single glycine receptor (GlyR). GlyRs have been found in various types of tissues, but their most important role seems to be in neurons, where they hyperpolarise membranes by opening chloride transmembrane channels. The software is based on a combination of two blocks. One block describes the Brownian dynamics of charged particle motion in a dielectric medium, and the other block determines the probability and timing of receptor activation. Using this software, the voltage-current dependencies and time curves of the transmembrane current were obtained. The mean value of the simulated anion current (4.5 +/- 0.3 pA) is in good agreement with measured values under identical conditions ( pA). It was shown that there is a condition under which the GlyR anion channel remains active despite a negligible chloride gradient. Virtual experiments allow evaluation of the value of half maximal effective concentration (EC) of the GlyR ( M) and confirm that this receptor activates according to a mechanism involving three ligand binding sites. The advantage of the model is the ability to adjust parameters to the precise demands of experimental researchers. Moreover, the introduced algorithm has low computational power demands;therefore, it can be used as a research tool for assistance with structural experiments and applied aspects of neurophysiology.

关键词： Membrane potential Modelling Ion currents stochastic algorithm

来源：评论

学校读者我要写书评

暂无评论

stochastic algorithmS FOR INVERSE PROBLEMS INVOLVING PDEs AND MANY MEASUREMENTS

引用

SIAM JOURNAL ON SCIENTIFIC COMPUTING 2014年第5期36卷 S3-S22页

作者： Roosta-Khorasani, Farbod van den Doel, Kees Ascher, Uri Univ British Columbia Dept Comp Sci Vancouver BC V6T 1W5 Canada

Inverse problems involving systems of partial differential equations (PDEs) can be very expensive to solve numerically. This is so especially when many experiments, involving different combinations of sources and receivers, are employed in order to obtain reconstructions of acceptable quality. The mere evaluation of a misfit function (the distance between predicted and observed data) often requires hundreds and thousands of PDE solves. This article develops and assesses dimensionality reduction methods, both stochastic and deterministic, to reduce this computational burden. We assume that all experiments share the same set of receivers and concentrate on methods for reducing the number of combinations of experiments, called simultaneous sources, that are used at each stabilized Gauss-Newton iteration. algorithms for controlling the number of such combined sources are proposed and justified. Evaluating the misfit approximately, except for the final verification for terminating the process, always involves random sampling. Methods for selecting the combined simultaneous sources, involving either random sampling or truncated SVD, are proposed and compared. Highly efficient variants of the resulting algorithms are identified, and their efficacy is demonstrated in the context of the famous DC resistivity and EIT problems. We present in detail our methods for solving such inverse problems. These methods involve incorporation of a priori information such as piecewise smoothness, bounds on the sought conductivity surface, or even a piecewise constant solution.

关键词： inverse problem stochastic algorithm partial differential equation many experiments DC resistivity EIT

来源：评论

学校读者我要写书评

暂无评论

Fast Kernel-based Method for Anomaly Detection

Fast Kernel-based Method for Anomaly Detection

引用

International Joint Conference on Neural Networks (IJCNN)

作者： Anh Le Trung Le Khanh Nguyen Van Nguyen Thai Hoang Le Dat Tran HCMc Univ Pedag Fac Informat Technol Ho Chi Minh City Vietnam VNUHCM Univ Sci Fac Informat Technol Ho Chi Minh City Vietnam Univ Canberra Fac Educ Sci Technol & Math Canberra ACT Australia

ISBN: (纸本)9781509006199

Anomaly detection (AD) involves detecting abnormality from normality and has a wide spectrum of applications in reality. Kernel-based methods for AD have been proven robust with diverse data distributions and offering good generalization ability. stochastic gradient descent (SGD) method has recently emerged as a promising framework to devise ultra-fast learning methods. In this paper, we conjoin the advantages of Kernel-based method and SGD-based method to propose fast learning methods for anomaly detection. We validate the proposed methods on 8 benchmark datasets in UCI repository and KDD cup 1999 dataset. The experimental results show that the proposed methods offer a comparable one-class classification accuracy while simultaneously achieving a significantly computational speed-up.

关键词： Anomaly detection Kernel method stochastic algorithm

来源：评论

学校读者我要写书评

暂无评论

stochastic QUASI-FEJER BLOCK-COORDINATE FIXED POINT ITERATIONS WITH RANDOM SWEEPING

引用

SIAM JOURNAL ON OPTIMIZATION 2015年第2期25卷 1221-1248页

作者： Combettes, Patrick L. Pesquet, Jean-Christophe Univ Paris 06 Sorbonne Univ UMR 7598 Lab Jacques Louis Lions F-75005 Paris France Univ Paris Est CNRS UMR 8049 Lab Informat Gaspard Monge F-77454 Marne La Vallee 2 France

This work proposes block-coordinate fixed point algorithms with applications to nonlinear analysis and optimization in Hilbert spaces. The asymptotic analysis relies on a notion of stochastic quasi-Fejer monotonicity, which is thoroughly investigated. The iterative methods under consideration feature random sweeping rules to select arbitrarily the blocks of variables that are activated over the course of the iterations and they allow for stochastic errors in the evaluation of the operators. algorithms using quasi-nonexpansive operators or compositions of averaged nonexpansive operators are constructed, and weak and strong convergence results are established for the sequences they generate. As a by-product, novel block-coordinate operator splitting methods are obtained for solving structured monotone inclusion and convex minimization problems. In particular, the proposed framework leads to random block-coordinate versions of the Douglas-Rachford and forward-backward algorithms and of some of their variants. In the standard case of m = 1 block, our results remain new as they incorporate stochastic perturbations.

关键词： arbitrary sampling block-coordinate algorithm fixed-point algorithm monotone operator splitting primal-dual algorithm stochastic quasi-Fejer sequence stochastic algorithm structured convex minimization problem

来源：评论

学校读者我要写书评

暂无评论

Increasing the efficiency of existing complex downstream processes by plant-wide retrofit

引用

CHEMICAL ENGINEERING RESEARCH & DESIGN 2015年 99卷 199-207页

作者： Lorenz, Hilke-Marie Otte, Daniel Staak, Daniel Repke, Jens-Uwe TU Bergahademie Freiberg Inst Thermal Environm & Nat Prod Proc Engn D-09596 Freiberg Germany Lonza Ltd CH-3930 Visp Switzerland

The economic feasibility of processes in the chemical industry is often determined by the energy and raw material demand, hence process efficiency is crucial. Most published methods to increase the efficiency of industrial processes are limited by a maximum possible degree of complexity of the respective processes. Thus, a systematic approach for the retrofit of existing processes is developed in the presented work, particularly for processes with industrial relevant complexity. The hybrid approach combines heuristic and mathematical methods. Due to recycle streams in the processes, retrofit modifications can affect the entire process, thus the operating points of all units have to be adapted. Therefore, a mathematical optimisation problem is formulated and a method based on the global optimisation algorithm molecular inspired parallel tempering (MIPT) is developed. In order to solve the optimisation problem, MATLAB (TM) and Chemcad (TM) are linked via an OPC interface. The developed method is successfully applied to an existing complex industrial world-scale process, which comprises a large number of components, including 18 unit operations and a considerable number of recycle streams. The presented retrofit leads to an increase of the efficiency of this complex process. (C) 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

关键词： Retrofit MINLP Multicomponent mixtures Global optimisation Sensitivity study stochastic algorithm

来源：评论

学校读者我要写书评

暂无评论

A sum-over-paths algorithm for third-order impulse-response moment extraction within RC IC-interconnect networks

引用

SOLID-STATE ELECTRONICS 2015年第Jul.期109卷 63-71页

作者： Wojcik, E. A. Ni, D. Lam, T. M. Le Coz, Y. L. Rensselaer Polytech Inst ECSE Dept Troy NY 12180 USA

We have created the first stochastic SoP (Sum-over-Paths) algorithm to extract third-order impulse-response (IR) moment within RC IC interconnects. It employs a newly discovered Feynman SoP Postulate. Importantly, our algorithm maintains computational efficiency and full parallelism. Our approach begins with generation of s-domain nodal-voltage equations. We then perform a Taylor-series expansion of the circuit transfer function. These expansions yield transition diagrams involving mathematical coupling constants, or weight factors, in integral powers of complex frequency s. Our SoP Postulate enables stochastic evaluation of path sums within the circuit transition diagram to order s(3)-corresponding to the order of IR moment (m(3)) we seek here. We furnish, for the first time, an informal algebraic proof independently validating our SoP Postulate and algorithm. We list, as well, detailed procedural steps, suitable for coding, that define an efficient stochastic algorithm for m3 IR extraction. Origins of the algorithm's statistical "capacitor-number cubed" correction and "double-counting" weight factors are explained, for completeness. Our algorithm was coded and successfully tested against exact analytical solutions for 3-, 5-, and 10-stage RC lines. We achieved better than 0.65% 1-sigma error convergence, after only 10K statistical samples, in less than 1 s of 2-GHz Pentium (R) execution time. These results continue to suggest that stochastic SoP algorithms may find useful application in circuit analysis of massively coupled networks, such as those encountered in high-end digital IC-interconnect CAD. (C) 2015 Elsevier Ltd. All rights reserved.

关键词： RC-circuits Moments Path sum stochastic algorithm

来源：评论

学校读者我要写书评

暂无评论

Unsupervised Nearest Neighbors Clustering With Application to Hyperspectral Images

引用

IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING 2015年第6期9卷 1105-1116页

作者： Cariou, Claude Chehdi, Kacem Univ Rennes 1 IETR TSI2M Team CNRS UMR 6164 F-22300 Lannion France

We address the problem of unsupervised clustering of multidimensional data when the number of clusters is not known a priori. The proposed iterative approach is a stochastic extension of the kNN density-based clustering (KNNCLUST) method which randomly assigns objects to clusters by sampling a posterior class label distribution. In our approach, contextual class-conditional distributions are estimated based on a k nearest neighbors graph, and are iteratively modified to account for current cluster labeling. Posterior probabilities are also slightly reinforced to accelerate convergence to a stationary labeling. A stopping criterion based on the measure of clustering entropy is defined thanks to the Kozachenko-Leonenko differential entropy estimator, computed from current class-conditional entropies. One major advantage of our approach relies in its ability to provide an estimate of the number of clusters present in the data set. The application of our approach to the clustering of real hyperspectral image data is considered. Our algorithm is compared with other unsupervised clustering approaches, namely affinity propagation (AP), KNNCLUST and Non Parametric stochastic Expectation Maximization (NPSEM), and is shown to improve the correct classification rate in most experiments.

关键词： Bayes' decision rule data clustering differential entropy estimation hyperspectral images nearest neighbors pixel classification stochastic algorithm

来源：评论

学校读者我要写书评

暂无评论

Bayesian Mixed Effect Atlas Estimation with a Diffeomorphic Deformation Model

引用

SIAM JOURNAL ON IMAGING SCIENCES 2015年第3期8卷 1367-1395页

作者： Allassonniere, S. Durrleman, S. Kuhn, E. Ecole Polytech CMAP F-91128 Palaiseau France INRIA Paris Roquencourt Aramis Lab F-75013 Paris France Univ Paris 06 Sorbonne Univ UMR 1127 F-75013 Paris France CNRS UMR 7225 F-75013 Paris France ICM F-75013 Paris France INRA Unite MIA F-78352 Jouy En Josas France

In this paper we introduce a diffeomorphic constraint on the deformations considered in the deformable Bayesian mixed effect template model. Our approach is built on a generic group of diffeomorphisms, which is parameterized by an arbitrary set of control point positions and momentum vectors. This enables us to estimate the optimal positions of control points together with a template image and parameters of the deformation distribution which compose the atlas. We propose to use a stochastic version of the expectation-maximization algorithm where the simulation is performed using the anisotropic Metropolis adjusted Langevin algorithm. We propose also an extension of the model including a sparsity constraint to select an optimal number of control points with relevant positions. Experiments are carried out on the United States Postal Service database, on mandibles of mice, and on three-dimensional murine dendrite spine images.

关键词： deformable template model atlas estimation diffeomorphic deformations stochastic algorithm anisotropic MALA control point optimization sparsity

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：