Abstract DEX is a qualitative multi-criteria decision analysis (MCDA) method, aimed at supporting decision makers in evaluating and choosing decision alternatives which has impact on security. We present results of a preliminary study in which we experimentally assessed the performance of two wellknown MCDA methods UTA and ACUTA to approximate qualitative DEX utility functions with piecewise-linear marginal utility functions. This is seen... DEX is a qualitative multi-criteria decision analysis (MCDA) method, aimed at supporting decision makers in evaluating and choosing decision alternatives which has impact on security. We present results of a preliminary study in which we experimentally assessed the performance of two wellknown MCDA methods UTA and ACUTA to approximate qualitative DEX utility functions with piecewise-linear marginal utility functions. This is seen as a way to improve the sensitivity of qualitative models and provide a better insight in DEX utility functions. The results indicate that the approach is in principle feasible, but at this stage suffers from problems of convergence, insufficient sensitivity and inappropriate handling of symmetric functions Read More Read Less

Abstract In surrogate-model-based optimization, the selection of an appropriate surrogate model is very important. If solution approximations returned by a surrogate model are accurate and with narrow confidence intervals, an algorithm using this surrogate model needs less exact solution evaluations to obtain results comparable to an algorithm without surrogate models. In this paper we compare two well known modeling techniques, random forest... In surrogate-model-based optimization, the selection of an appropriate surrogate model is very important. If solution approximations returned by a surrogate model are accurate and with narrow confidence intervals, an algorithm using this surrogate model needs less exact solution evaluations to obtain results comparable to an algorithm without surrogate models. In this paper we compare two well known modeling techniques, random forest (RF) and Gaussian process (GP) modeling. The comparison includes the approximation accuracy and confidence in the approximations (expressed as the confidence interval width). The results show that GP outperforms RF and that it is more suitable for use in a surrogatemodel- based multiobjective evolutionary algorithm. Read More Read Less

Abstract One of the basic problems that cause loss of yield in wheat is weed seeds that mixed with wheat seeds. In this study, discrimination of barley seed which mixed with wheat seeds has been realized. Classification of wheat and barley seeds has been achieved by using artificial neural network and image processing techniques. In the study, image processing techniques and the use of... One of the basic problems that cause loss of yield in wheat is weed seeds that mixed with wheat seeds. In this study, discrimination of barley seed which mixed with wheat seeds has been realized. Classification of wheat and barley seeds has been achieved by using artificial neural network and image processing techniques. In the study, image processing techniques and the use of artificial neural network have been made possible with Matlab software. By using Otsu method, histogram data of seed images that were taken from web camera was obtained. By using histogram data, with multi-layered artificial neural network model, the system was educated and classification was made. Besides, wheat and barley seeds in the picture info where mixed seeds taken from the web camera exist were counted. Read More Read Less

Abstract Comprehensibility is the decisive factor for application of classifiers in practice. However, most algorithms that learn comprehensible classifiers use classification model size as a metric that guides the search in the space of all possible classifiers instead of comprehensibility - which is ill-defined. Several surveys have shown that such simple complexity metrics do not correspond well to the comprehensibility of classification trees. This... Comprehensibility is the decisive factor for application of classifiers in practice. However, most algorithms that learn comprehensible classifiers use classification model size as a metric that guides the search in the space of all possible classifiers instead of comprehensibility - which is ill-defined. Several surveys have shown that such simple complexity metrics do not correspond well to the comprehensibility of classification trees. This paper therefore suggests a classification tree comprehensibility survey in order to derive an exhaustive comprehensibility metrics better reflecting the human sense of classifier comprehensibility and obtain new insights about comprehensibility of classification trees. Read More Read Less