In this paper, we propose a trust region method for solving KKT systems arising from the variational inequality problem and the constrained optimization problem. The trust region subproblem is derived from reformulation of the KKT system as a constrained optimization problem and is solved by the truncated conjugate gradient method; meanwhile the variables remain feasible with respect to the con...

This article presents a constrained optimization method, based on the duality theory, which does not need the gradients. The method is used to optimize superconducting devices. In order to reduce the computing effort, the initial optimization problem is divided into two coupled optimization problems. One manages the geometrical parameters, the other finds the best current densities for a given ...

Cuckoo search (CS) has been recently proposed as a population-based optimization algorithm and it is has so far been successfully applied in a variety of fields. An efficient hybrid cuckoo search algorithm (HCSA) based on Powell direct search method is proposed for solving constrained engineering design optimization problems. The inertia weight of Levy flights is introduced to balance the abili...

In this paper we propose a simple adaptive algorithm iteratively solving the unit-norm constrained optimization problem. Instead of conventional parameter norm based normalization, the proposed algorithm incorporates scalar normalization which is computationally much simpler. The analysis of stationary point is presented to show that the proposed algorithm indeed solves the constrained optimiza...

In this Chapter, we present a new face detection and tracking algorithm using Bayesconstrained particle swarm optimization (BC-PSO), which is a population based searching algorithm. A cascade of boosted classifiers based on Haar-like features is trained and employed for object detection. Then the PSO-based algorithm is applied for object tracking. Basically the searching can be divided into two...

For many applications of reinforcement learning it can be more convenient to specify both a reward function and constraints, rather than trying to design behavior through the reward function. For example, systems that physically interact with or around humans should satisfy safety constraints. Recent advances in policy search algorithms (Mnih et al., 2016; Schulman et al., 2015; Lillicrap et al...

All Rights Reserved ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy. I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy. I certify that I have read this di...

Many optimization models of neural networks need constraints to restrict the space of outputs to a subspace which satisfies external criteria. Optimizations using energy methods yield "forces" which act upon the state of the neural network. The penalty method, in which quadratic energy constraints are added to an existing optimization energy, has become popular recently, but is not guaranteed t...

EQUALITY CONSTRAINED OPTIMIZATION Masao Fukushima Kyoto University Hisashi Mine Kansai University Eiki Yamakawa Kyoto University (Received November 12, 1984: Revised June 12, 1985) This paper is concerned with a differentiable exact penalty function derived by modifying the Wolfe dual of an equality constrained problem. It may be considered that this penalty function belongs to a class of gener...