Asymptotic behavior of the online gradient algorithm with a constant step size employed for learning in neural network models of nonlinear systems having hidden layer are studied. The sufficient conditions guaranteein...
详细信息
ISBN:
(纸本)9781479933068;9781479933051
Asymptotic behavior of the online gradient algorithm with a constant step size employed for learning in neural network models of nonlinear systems having hidden layer are studied. The sufficient conditions guaranteeing the convergence of this algorithm in the random environment are established.
The concave utilities in the basic network utility maximisation (NUM) problem are only suitable for elastic flows. In networks with both elastic and inelastic traffic, the utilities of inelastic traffic are usually mo...
详细信息
The concave utilities in the basic network utility maximisation (NUM) problem are only suitable for elastic flows. In networks with both elastic and inelastic traffic, the utilities of inelastic traffic are usually modelled by the sigmoidal functions which are non-concave functions. Hence, the basic NUM problem becomes a non-convex optimisation problem. To address the non-convex NUM, the authors approximate the problem which is equivalent to the original one to a strictly convex problem. The approximation problem is solved efficiently via its dual by the gradient algorithm. After a series of approximations, the sequence of solutions to the approximation problems converges to a local optimal solution satisfying the Karush-Kuhn-Tucker conditions of the original problem. The proposed algorithm converges with any value of link capacity. The authors also extend their work to jointly allocate the rate and the power in a multihop wireless network with elastic and inelastic traffic. Their framework can be used for any log-concave utilities.
For the calculation of gravity and magnetic source boundaries, the traditional gradient methods are vulnerable to disturbance and the resulting edges may be too cluttered when the field data are affected by noise. Bes...
详细信息
For the calculation of gravity and magnetic source boundaries, the traditional gradient methods are vulnerable to disturbance and the resulting edges may be too cluttered when the field data are affected by noise. Besides, for the weak anomaly, sources tend to interfere with one another and consequently source edges may be impossible to detect. In this paper, we first construct an anisotropic Gauss function based on coordinate rotation, then we propose the formula of calculating anisotropy normalized variance to detect source edges, and clarify its geological significance through theoretical analysis. The numerical experiment shows the method is stable and effective. Applications to both synthetic data and aeromagnetic data in mid Yangtze region indicate that calculation of anisotropy normalized variance can effectively determine the gravity and magnetic source boundaries, especially for weak anomalies. The calculated boundaries bear high-resolution and ample information, and will facilitate the integrated interpretation of data.
In this paper, a penalty term is added to the conventional error function to improve the generalization of the Ridge Polynomial neural network. In order to choose appropriate learning parameters, we propose a monotoni...
详细信息
In this paper, a penalty term is added to the conventional error function to improve the generalization of the Ridge Polynomial neural network. In order to choose appropriate learning parameters, we propose a monotonicity theorem and two convergence theorems including a weak convergence and a strong convergence for the synchronous gradient method with penalty for the neural network. The experimental results of the function approximation problem illustrate the above theoretical results are valid. (C) 2012 Elsevier B.V. All rights reserved.
作者：
Penenko, A. V.Russian Acad Sci
Inst Computat Math & Math Geophys Siberian Branch Pr Akad Lavrenteva 6 Novosibirsk 630090 Russia
A method for constructing numerical schemes for an inverse coefficient heat conduction problem with boundary measurement data and piecewise-constant coefficients is considered. Some numerical schemes for a gradient op...
详细信息
A method for constructing numerical schemes for an inverse coefficient heat conduction problem with boundary measurement data and piecewise-constant coefficients is considered. Some numerical schemes for a gradient optimization algorithm to solve the inverse problem are presented. The method is based on locally-adjoint problems in combination with approximation methods in Hilbert spaces.
This article considers the time-dependent optimal control problem of tracking the velocity for the viscous incompressible flows which is governed by a Ladyzhenskaya equations with distributed control. The existence of...
详细信息
This article considers the time-dependent optimal control problem of tracking the velocity for the viscous incompressible flows which is governed by a Ladyzhenskaya equations with distributed control. The existence of the optimal solution is shown and the first-order optimality condition is established. The semidiscrete-in-time approximation of the optimal control problem is also given. The spatial discretization of the optimal control problem is accomplished by using a new stabilized finite element method which does not need a stabilization parameter or calculation of high order derivatives. Finally a gradient algorithm for the fully discrete optimal control problem is effectively proposed and implemented with some numerical examples. (C) 2010 Wiley Periodicals, Inc. Numer Methods Partial Differential Eq 28: 263-287, 2012
This paper provides new insights into methods performing automatic white balance for a digitally captured image. It is shown that automatic white balance may be formulated as an optimization problem with explicit defi...
详细信息
This paper provides new insights into methods performing automatic white balance for a digitally captured image. It is shown that automatic white balance may be formulated as an optimization problem with explicit definition of objective function, decision variables, and constraints. Three alternative methods of formulating the optimization problem are proposed. It is also shown that fuzzy inference rules, commonly utilized in existing literatures to evaluate to what degree an image satisfying the gray world assumption, may be incorporated into the objective function of the optimization problem. A two-stage adjustment law with constrained search direction is then proposed to update the decision variables. A gradient descent algorithm is employed to numerically solve the problem, which guarantees the convergence and that optimal white balance effort is achieved for most images. Experimental results and a comparative study justify that the proposed methods are preferable to existing methods with regard to the execution time, the algorithmic complexity, and the performance. (C) 2009 Elsevier B.V. All rights reserved.
Genetic arithmetic operators in genetic algorithm be improved, and a hybrid genetic algorithm of a gradient algorithm combining with the genetic algorithm be given against to the defects such as premature, slow on con...
详细信息
ISBN:
(纸本)9780878492060
Genetic arithmetic operators in genetic algorithm be improved, and a hybrid genetic algorithm of a gradient algorithm combining with the genetic algorithm be given against to the defects such as premature, slow on convergence rate, weak in the ability of local search, all these appeared on the progress of genetic algorithm's iteration. Analysis result indicate that not only strong on the local search capacity of gradient algorithm be exhibited but also strong on the general search capacity of genetic algorithm be combined based on the hybrid genetic algorithm,which make phenomenon of premature avoid, and the rate of convergence be improved greatly. Concrete calculated example indicated that the hybrid genetic algorithm is an effective structural optimization method.
A method is proposed for constructing noise-resistant mask differentiating filters minimizing systematic errors in estimating directional fields by a gradient algorithm. A comparative analysis of the accuracy of gradi...
详细信息
A method is proposed for constructing noise-resistant mask differentiating filters minimizing systematic errors in estimating directional fields by a gradient algorithm. A comparative analysis of the accuracy of gradient algorithms was performed by computer modeling It is shown that a decrease in the systematic error leads to a considerable increase in the accuracy of directional field estimation
This paper investigates shape optimization of a solid body located in Navier-Stokes flow in two dimensions. The minimization problem of total dissipated energy is established in the fluid domain. The discretization of...
详细信息
This paper investigates shape optimization of a solid body located in Navier-Stokes flow in two dimensions. The minimization problem of total dissipated energy is established in the fluid domain. The discretization of Navier-Stokes equations is accomplished using a new stabilized finite element method which does not need a stabilization parameter or calculation of high order derivatives. We derive the structures of discrete Eulerian derivative of the cost functional by a discrete adjoint method with a function space parametrization technique. A gradient type optimization algorithm with a mesh adaptation technique and a mesh moving strategy is effectively formulated and implemented. (C) 2010 Published by Elsevier B.V. on behalf of IMACS.
暂无评论