Search results for "basis"
showing 10 items of 760 documents
On Computational Properties of a Posteriori Error Estimates Based upon the Method of Duality Error Majorants
2004
In the present paper, we analyze computational properties of the functional type a posteriori error estimates that have been derived for elliptic type boundary-value problems by duality theory in calculus of variations. We are concerned with the ability of this type of a posteriori estimates to provide accurate upper bounds of global errors and properly indicate the distribution of local ones. These questions were analyzed on a series of boundary-value problems for linear elliptic operators of 2nd and 4th order. The theoretical results are confirmed by numerical tests in which the duality error majorant for the classical diffusion problem is compared with the standard error indicator used i…
Kernelizing LSPE(λ)
2007
We propose the use of kernel-based methods as underlying function approximator in the least-squares based policy evaluation framework of LSPE(λ) and LSTD(λ). In particular we present the 'kernelization' of model-free LSPE(λ). The 'kernelization' is computationally made possible by using the subset of regressors approximation, which approximates the kernel using a vastly reduced number of basis functions. The core of our proposed solution is an efficient recursive implementation with automatic supervised selection of the relevant basis functions. The LSPE method is well-suited for optimistic policy iteration and can thus be used in the context of online reinforcement learning. We use the hig…
Approximate survival probability determination of hysteretic systems with fractional derivative elements
2018
Abstract A Galerkin scheme-based approach is developed for determining the survival probability and first-passage probability of a randomly excited hysteretic systems endowed with fractional derivative elements. Specifically, by employing a combination of statistical linearization and of stochastic averaging, the amplitude of the system response is modeled as one-dimensional Markovian Process. In this manner the corresponding backward Kolmogorov equation which governs the evolution of the survival probability of the system is determined. An approximate solution of this equation is sought by employing a Galerkin scheme in which a convenient set of confluent hypergeometric functions is used a…
Adaptive Gaussian particle method for the solution of the Fokker-Planck equation
2012
The Fokker-Planck equation describes the evolution of the probability density for a stochastic ordinary differential equation (SODE). A solution strategy for this partial differential equation (PDE) up to a relatively large number of dimensions is based on particle methods using Gaussians as basis functions. An initial probability density is decomposed into a sum of multivariate normal distributions and these are propagated according to the SODE. The decomposition as well as the propagation is subject to possibly large numeric errors due to the difficulty to control the spatial residual over the whole domain. In this paper a new particle method is derived, which allows a deterministic error…
Solving continuous models with dependent uncertainty: a computational approach
2013
This paper presents a computational study on a quasi-Galerkin projection-based method to deal with a class of systems of random ordinary differential equations (r.o.d.e.'s) which is assumed to depend on a finite number of random variables (r.v.'s). This class of systems of r.o.d.e.'s appears in different areas, particularly in epidemiology modelling. In contrast with the other available Galerkin-based techniques, such as the generalized Polynomial Chaos, the proposed method expands the solution directly in terms of the random inputs rather than auxiliary r.v.'s. Theoretically, Galerkin projection-based methods take advantage of orthogonality with the aim of simplifying the involved computat…
The Power of the “Pursuit” Learning Paradigm in the Partitioning of Data
2019
Traditional Learning Automata (LA) work with the understanding that the actions are chosen purely based on the “state” in which the machine is. This modus operandus completely ignores any estimation of the Random Environment’s (RE’s) (specified as \(\mathbb {E}\)) reward/penalty probabilities. To take these into consideration, Estimator/Pursuit LA utilize “cheap” estimates of the Environment’s reward probabilities to make them converge by an order of magnitude faster. This concept is quite simply the following: Inexpensive estimates of the reward probabilities can be used to rank the actions. Thereafter, when the action probability vector has to be updated, it is done not on the basis of th…
Representation and estimation of spectral reflectances using projection on PCA and wavelet bases
2008
In this article, we deal with the problem of spectral reflectance function representation and estimation in the context of multispectral imaging. Because the reconstruction of such functions is an inverse problem, slight variations in input data completely skew the expected results. Therefore, stabilizing the reconstruction process is necessary. To do this, we propose to use wavelets as basis functions, and we compare those with Fourier and PCA bases. We present the idea and compare these three methods, which belong to the class of linear models. The PCA method is training-set dependent and confirms its robustness when applied to reflectance estimation of the training sets. Fourier and wave…
An Introduction to Kernel Methods
2009
Machine learning has experienced a great advance in the eighties and nineties due to the active research in artificial neural networks and adaptive systems. These tools have demonstrated good results in many real applications, since neither a priori knowledge about the distribution of the available data nor the relationships among the independent variables should be necessarily assumed. Overfitting due to reduced training data sets is controlled by means of a regularized functional which minimizes the complexity of the machine. Working with high dimensional input spaces is no longer a problem thanks to the use of kernel methods. Such methods also provide us with new ways to interpret the cl…
Designing Paper Machine Headbox Using GA
2003
Abstract A non-smooth biobjective optimization problem for designing the shape of a slice channel in a paper machine headbox is described. The conflicting goals defining the optimization problem are the ones determining important quality properties of produced paper: 1) basis weight should be even and 2) the wood fibers of paper should mainly be oriented to the machine direction across the width of the whole paper machine. The novelty of the considered approach is that maximum deviations are used instead of least squares when objective functions are formed. For the solution of this problem, a multiobjective genetic algorithm based on nondominated sorting is considered. The numerical results…
Heat semi-group and generalized flows on complete Riemannian manifolds
2011
Abstract We will use the heat semi-group to regularize functions and vector fields on Riemannian manifolds in order to develop Di Perna–Lions theory in this setting. Malliavinʼs point of view of the bundle of orthonormal frames on Brownian motions will play a fundamental role. As a byproduct we will construct diffusion processes associated to an elliptic operator with singular drift.