How Generalised Singular Vectors Can Help to Develop New Regularisation Methods

来源 :第八届工业与应用数学国际大会 | 被引量 : 0次 | 上传用户：tongjingjj

【摘要】

：

　　Singular value decomposition is the key tool in the analysis and understanding of linear regularisation methods.Recently,the notion and concept of singular

【作者】

：

BenningMartin;

【机构】

：

Univ.of Cambridge

【出处】

：

第八届工业与应用数学国际大会

【发表日期】

：

2015年期

下载到本地 , 更方便阅读

下载此文赞助VIP

声明 : 本文档内容版权归属内容提供方 , 如果您对本文有版权争议 , 可与客服联系进行内容授权或下架

论文部分内容阅读

　　Singular value decomposition is the key tool in the analysis and understanding of linear regularisation methods.Recently,the notion and concept of singular vectors has been generalised to non-linear regularisation methods like l1-or total variation-regularisation.

其他文献

Initiative Taken by Institute of Mathematics for Industry,Kyushu University

　　Mathematics for Industry(MI)has been born by amalgamating and reorganizing pure and applied mathematics,to serve as the foundation for future technologies.V

会议

Mathematics for Industry 1:Analytical,geometrical and statistical methods to solids,fluids and plasm

　　Mathematics for Industry(MI)has been born by amalgamating and reorganizing pure and applied mathematics,to serve as the foundation for developing future tec

会议

Modeling for Droplet Motion Driven by Interfacial Tension Gradient

　　It is known that a droplet with surface tension gradient can deform and move through the Marangoni effect.We constructed a mathematical model for such dropl

会议

The String Method for Saddle Point Search

　　In this talk,we show how the string method,which was originally developed to compute minimum energy paths between two meta-stable states,can be used to comp

会议

Evolving Junctions on Obstacle Boundaries(E-JOB)Method for the Shortest Path Problems

　　In this talk,I will use the shortest path problem,finding the shortest path connecting two points while avoiding obstacles in a region,as an example to illu

会议

An Adaptive Step-size String Method and Its Convergence to the Minimum Energy Path

　　Finding minimum energy path(MEP)on a potential energy surface is of great interest in understanding the barrier-crossing events.We develop an adaptive step-

会议

Variational Nonlinear Eigenfunction Analysis for Signal Representation and Processing

　　Linear eigenfunction decomposition is a powerful technique used broadly for processing signals and images.In this talk a generalization based on convex vari

会议

Derivative-free Nonlinear Constrained Optimization under Uncertainty Using NOWPAC

　　We present the derivative-free optimization procedure NOWPAC(Nonlinear Optimization With Path-Augmented Constraints)for nonlinear constrained programming.Th

会议

Metastability,Spectra and Eigencurrents of Networks Representing Energy Landscapes

　　Computational tools for spectral analysis of large stochastic networks representing energy landscapes of atomic and molecular clusters are proposed:(i)an ef

会议

Designing A Realistic Image Denoising Model by Means of A Bilevel Optimization Approach

　　We consider a nonlinear PDE-constrained optimization approach to learn the optimal weights for a total variation image denoising model featuring different n

会议

How Generalised Singular Vectors Can Help to Develop New Regularisation Methods

其他学术论文