Marketing decision making tasks require the acquisition of efficient decision rules from noisy questionnaire data. Unlike popular learning-from-example methods, in such tasks, we must interpret the characteristics of the data without clear features of the data nor pre-determined evaluation criteria. The problem is how domain experts get simple, easy-to-understand, and accurate knowledge from noisy data.
This paper describes a novel method to acquire efficient decision rules from questionnaire data using both simulated breeding and inductive learning techniques. The basic ideas of the method are that simulated breeding is used to get the effective features from the questionnaire data and that inductive learning is used to acquire simple decision rules from the data. The simulated breeding is one of the Genetic Algorithm based techniques to subjectively or interactively evaluate the qualities of offspring generated by genetic operations.
The proposed method has been qualitatively and quantitatively validated by a case study on consumer product questionnaire data: the acquired rules are simpler than the results from the direct application of inductive learning; a domain expert admits that they are easy to understand; and they are at the same level on the accuracy compared with the other methods.
Furthermore, we address three variations of the basic interactive version of the method: (i) with semiautomated GA phases, (ii) with the relatively evaluation phase via AHP, and (iii) with an automated multiagent learning method. 相似文献
One of the crucial issues in engineering problem solving is prediction of behavior, namely deriving the behavior from models. Prediction of behavior is not just calculation. Rather, it is an intellectual activity. High level decision making is needed to plan what and how to compute, monitor computation process, and interpret the result. Hence, a large portion of the prediction process has been indebted to human experts. As a first step towards this end, we concentrate on systems of piecewise linear ordinary differential equations (systems of PWL-ODEs), and proposeFlow Topology Analysis, a computational framework of analyzing systems of PWL-ODEs. We propose abstract representation, calledflow mappings, for describing qualitative aspects of solution curves in regions of the phase space. Flow mappings are used as an interface between knowledge and numerical levels. Currently, we have implemented the idea for systems of PWL-ODEs on two-dimensional phase spaces and partly for those on three-dimensional phase spaces. 相似文献
Multimedia Tools and Applications - The difficulties and social anxiety associated with living in unfamiliar places are often caused by different patterns of thinking, points of view (POVs) and... 相似文献
A gyroscopic sensor using active magnetic bearing (AMB) is studied both theoretically and experimentally. The sensor has been proposed to realize high accuracy, compact and low-cost sensors, and it utilizes the control function of the AMB. The sensor works as a two-axis gyroscopic sensor and also as a three-axis servo-type accelerometer. Angular velocities and accelerations are measured based on the control signals for cancelling the inertial effects that act on the AMB rotor. The methods to detect the accelerations and angular velocities have been validated from several experimental results. In addition, it has been indicated that there is an upper limit in the measurement bandwidth of the sensor when the two-axis angular velocity is measured simultaneously. However, factors determining the limit have not been presented in a concrete manner. This paper discusses these factors and presents a methodology that can extend the measurement bandwidth of the sensor. The limiting factors are investigated extensively by analysis, numerical simulations and experiments with the AMB. The relationship between the measurement bandwidth and the control system is also investigated. In addition, the measurement range of the gyro is discussed. 相似文献
1.5-μm wavelength strained-layer multiple-quantum-well (SL-MQW) distributed-feedback (DFB) lasers for optical video distribution systems, including optical-fiber amplifiers, are studied with respect to the relaxation-oscillation frequency and wavelength chirp characteristics. Several types of lasers are examined as parameters of amount of strain, optical confinement, and detuning. It is confirmed that the introduction of negative detuning has an obvious effect on the increase in relaxation-oscillation frequency fr and the reduction in chirp ΔF. The SL-MQW DFB lasers with low optical confinement and negative detuning show extremely low FM response ΔF/ΔI of less than 60 MHz/mA as well as high fr over 10 GHz for the first time. Additionally, low-chirp value ΔF of 180 MHz with modulation depth of 10% is realized at relatively low bias current (Ih=Ith+30 mA) 相似文献
Two solid solutions EuxSr1?xNb4O11 and EuxSr1?xTa4O11 have been prepared. Electrical and magnetic properties of these solid solutions were investigated. Electrical conductivity in the niobate system was found to be much higher than that in the tantalate analog. The specimen Eu0.1Sr0.9Nb4O11 was most transmitting in the solid solutions in the range of 77–450K. 相似文献
We present an approach based on knowledge medium using associative structures as a framework of information representation to gather raw information from heterogeneous information sources and to integrate it into information bases cost-effectively.We then present a knowledge media information base system called CM-2 which provides users with a means of accumulating, sharing, exploring and refining conceptually diverse information gathered from vast information sources. We describe the system's four major facilities; (a) an information capture facility, (b) an information integration facility, (c) an information retrieval facility and (d) an information refinement facility. We discuss the strength and weakness of our approach by analyzing results of experiments. 相似文献