We improve the efficiency of a DMU considering its interval efficiency value based on DEA by adjusting its outputs. The DEA model to obtain an interval efficiency value which consists of evaluations from both the optimistic and pessimistic viewpoints has been formulated. DMUs which are not rated as efficient in the conventional sense are improved by adjusting outputs so that their lower bounds become as large as possible in the production possibility set and their upper bounds attain the maximum value one. The lower bound of interval efficiency value becomes small when the DMU's inputs and outputs are not balanced. Therefore considering the lower bounds in improvement, the adjusted outputs can keep the balance each other. A new approach to improvement by adjusting only outputs without decreasing the given outputs anymore is proposed in this paper. By extending the proposed approach, it is suggested that the approaches to improvement by adjusting only inputs and both inputs and outputs simultaneously can be obtained. Numerical examples are shown to illustrate our proposed approaches.

View full abstract

We attempted health and disease state estimation with data-mining using Self-Organizing Maps (SOM) for application to plethysmogram information. Such information is easily gained from patient fingertip sensors. We used eight variables, such as chaotic analysis values calculated by the trajectory parallel measure method, and the recurrence plot method, in addition to the waveform component ratio, which is a linear analysis value of acceleration plethysmogram. As conventional studies have reported, SOM also confirmed that the waveform component ratio is related to aging. Self-organized acceleration plethysmogram information showed that trajectory parallel measure method values such as chaotic analysis values are useful for disease state estimation of circulatory failure, arteriosclerosis, or acute inflammation. Moreover, the recurrence plot method may reflect the presence and gravity of a disease state. This study, rather than offering general diagnoses of topical named diseases, suggests the possibility of health evaluation by the blood flow state.

View full abstract

This paper presents a self-tuning system of PID gain parameters for an inverted pendulum control system using three-layer neural networks. The inverted pendulum system which has one input and two output system is expressed as the plant by the transfer functions, which are used to identify the plant with a neural network based on the back-propagation for temporal sequences and then the system Jacobian of the plant is derived. The system Jacobian of the plant is used in the self-tuning process of the PID controller. The PID parameters are determined so as to minimize the error function, in which another three-layer neural network is used in the tuning process. Experimental results of the angle of the pendulum and the position of the cart which are controlled by the tuned parameteres are compared with simulation results. It is shown that they are good agreement.

View full abstract

In this paper, we focus on multiobjective integer programming problems with random variable coefficients in objective functions and/or constraints. For such multiobjective problems, after reformulation of them on the basis of an expectation optimization model and a variance minimization model for the chance constrained programming in stochastic programming, incorporating fuzzy goals of the decision maker for the objective functions, we propose an interactive fuzzy satisficing method to derive a satisficing solution for the decision maker as a fusion of stochastic programming and fuzzy one. Since some nonliner 0-1 programming problems must be solved in the proposed method, we apply a genetic algorithm with double string using reference solution updating.

View full abstract

In a multi-agent system as a simulation model of social systems, the equilibrium of the system is attained at the competitive equilibrium solution which satisfies the individual rationality. On the other hand, since an agent as a model of a human being usually belongs to some group, it may be often desirable that the system has a cooperative equilibrium solution which satisfies the group rationality. Furthermore, information which an agent uses for the selection of actions is supposed to be accurate in most of past researches though it seems more practical that it includes uncertainty and delay. In this paper, focusing on Hogg-Huberman model originally composed of only competitive agents with the uncertain and delayed information, we introduce another type of agent based on the group rationality. As a result, we realize a cooperative equilibrium to maxmize the total profit of the system. Then, through a number of computer simulations, we investigate and discuss the stability of a system which consists of either competitive agents or cooperative agents together with the influence of the ratio of cooperative agents in the system and the degree of information uncertainty on the total profit of the system.

View full abstract