It is shown that, in comparison to the results obtained from a conventional least squares approach, a total least squares solution leads to significant improvements in the geometry and appearance of images synthesised in a linear combination of views procedure. Use of the total least squares criterion is appropriate when errors on the control points are independently and identically distributed...

A method for the construction of open approximate models from vector time series Preface PhD research is a largely open proces in which a stimulating environment plays a crucial role. I am indebted to several people and institutions that shaped such an environment for me during the years since April 1990. First and for all, I would like to thank my supervisor, Christiaan Heij. He set me on the ...

We formulate the problem of least squares temporal difference learning (LSTD) in the framework of least squares SVM (LS-SVM). To cope with the large amount (and possible sequential nature) of training data arising in reinforcement learning we employ a subspace based variant of LS-SVM that sequentially processes the data and is hence especially suited for online learning. This approach is adapte...

Optimal control problems with convex functions are ubiquitous in applications of stochastic optimization. However, when applied in this context, the classical least squares Monte Carlo methodology makes no attempt to take advantage of this special structure: Given the convexity of value functions, it seems reasonable to search for the best least-squares fit among the elements of a cone of conve...

Purpose: preliminary discussion on model prediction precision in the partial least squares regression analysis method; Method: introduce current development conditions of partial least squares regression analysis, analyze problems of traditional regression analysis method such as multiple linear regression analysis, introduce the mathematic principle and modeling method of the partial least squ...

We present a new algorithm for approximate joint diagonalization of several symmetric matrices. While it is based on the classical least squares criterion, a novel intrinsic scale constraint leads to a simple and easily parallelizable algorithm, called LSDIC (Least squares Diagonalization under an Intrinsic Constraint). Numerical simulations show that the algorithm behaves well as compared to o...