The handling of many-objective problems is a hot issue in the evolutionary multiobjective optimization (EMO) community. Whereas Pareto-based EMO algorithms usually work very well on two-objective problems, they do not work well on many-objective problems. A promising approach to the search for the non-dominated solutions of many-objective problems is a class of indicator-based EMO algorithms. The goal of indicator-based EMO algorithms is to maximize an indicator function which evaluates the quality of a set of solutions. The hypervolume has been frequently used as an indicator function. The main difficulty of the use of hypervolume is that the computation load for its calculation increases exponentially with the number of objectives. Thus the application of indicator-based EMO algorithms to many-objective problems is time-consuming. In our former study, we proposed an idea of approximating the hypervolume using a number of achievement functions with uniformly distributed weight vectors. In this paper, we incorporate our hypervolume approximation into indicator-based EMO algorithms. Experimental results show that the computation time of indicator-based EMO algorithms for many-objective problems is drastically decreased by the use of our hypervolume approximation method with no severe deterioration in its search ability.
In this paper, we propose a new selection scheme of initial solutions for the local search of a multiobjective genetic local search (MOGLS) algorithm. The MOGLS algorithm is the hybridization of an evolutionary multiobjective optimization (EMO) algorithm and local search. It is shown that the MOGLS algorithm has higher search ability than pure EMO algorithms. In the conventional MOGLS algorithm, the local search method is applied to the offspring population generated by the genetic operators. However, the generated offspring population often includes poor individuals because the genetic operators involve some random procedures and allow the generation of inferior offspring. The basic idea of our approach is to apply local search to the parent population. Thus our approach can apply local search to better solutions than the original MOGLS algorithm on average. Through computational experiments, we show that our approach improves the search ability of the MOGLS algorithm.
The button-press task means that the subject observes a moving target and presses a button to stop it when the target enters a specified area on a computer display. Subjects perform normal task, suppressed task and delayed task. In the suppressed task, the moving target disappears at some point during the trial. In the delayed task, there is some lag time between the time of pressing button and of stopping target. In these tasks, subjects estimate the movement of the target, and press the button considering his/her own reaction time. In our previous study, we showed that cognitive and motor function was able to be evaluated by these tasks. In this study, we examined error data of children with developmental disabilities to evaluate the cognitive function, and investigated the learning processes. Moreover, we discussed the developmental stages by comparing the children with disabilities to normal control children, and we clarified the behavior characteristics of children with developmental disabilities. Asa result, it was shown that our evaluation method and system for the button-press task were effective to evaluate cognitive ability of children with developmental disabilities.