There are several data based methods in the field of artificial intelligence which are nowadays frequently used for analyzing classification problems in the context of medical applications. As we show in this paper, the application of enhanced evolutionary computation techniques to classification problems has the potential to evolve classifiers of even higher quality than those trained by standard machine learning methods. On the basis of three medical benchmark classification problems, namely the Wisconsin and the Thyroid data sets taken from the UCI repository as well as the Melanoma data set prepared by members of the Department of Dermatology of the Medical University Vienna, we document that the enhanced genetic programming based approach presented here is able to produce better results than linear modeling methods, artificial neural networks, kNN classification and also standard genetic programming approaches.

In this paper the impact of nonlinear distortions on the linear system identification framework is studied. In the first part the nonlinear system is replaced by a linear model plus a nonlinear noise source. The properties of this representation are studied. Next a method to detect, qualify and quantify the nonlinear distortions is presented. In the second part, the (non)-parametric identification of the best linear approximation is studied. In the last part, the linear modelling approach is extended towards nonlinear modelling. A fast approximate nonlinear modelling framework is set up that is a natural extension of the linear framework, and bridges the gap between the linear and the nonlinear identification approaches.

In this work, by using the local node refinement technique purposed in [2, 1], and a quad-tree type algorithm [3, 13], we built a global refinement technique for Kansa's unsymmetric collocation approach. The proposed scheme is based on a cell by cell data structure, which by using the former local error estimator, iteratively refines the node density in regions with insufficient accuracy. We test our algorithm for steady state partial differential equations in one and two dimensions. By using thin-plate spline kernel functions, we found that the node refinement let us to reduce the approximation error and that the node insertion is only performed in regions where the analytical solution shows a high spatial variation. In addition, we found that the node refinement outperform in accuracy and number of nodes in comparison with the global classical Cartesian h- refinement technique.

A Bayesian method for the comparison and selection of multi-output feedforward neural network topology, based on the predictive capability, is proposed . As a measure of the prediction fitness potential, an expected utility criterion is considered which is consistently estimated by a sample-reuse computation. As opposed to classic point-prediction-based crossvalidation methods, this expected utility is defined from the logarithmic score of the neural model predictive probability density. It is shown how the advocated choice of a conjugate probability distribution as prior for the parameters of a competing network, allows a consistent approximation of the network posterior predictive density. A comparison of the performances of the proposed method with the performances of usual selection procedures based on classic cross-validation and informationtheoretic criteria, is performed first on a simulated case study, and then on a well-known food analysis dataset.

This paper presents a novel Wiener-type recurrent neural network with the observer/Kalman filter identification (OKID) algorithm for unknown dynamic nonlinear system identification. The proposed Wiener-type recurrent network resembles the conventional Wiener model that consists of a dynamic linear subsystem cascaded with a static nonlinear subsystem. The novelties of our approach include: (1) the realization of a conventional Wiener model into a simple connectionist recurrent network whose output can be expressed by a nonlinear transformation of a linear state-space equation; (2) the overall network structure can be determined by the OKID algorithm effectively using only the input-output measurements; and (3) the proposed network is capable of accurately identifying nonlinear dynamic systems using fewer parameters. Computer simulations and comparisons with some existing recurrent networks and learning algorithms have successfully confirmed the effectiveness and superiority of the proposed Wienertype network with the OKID algorithm.

other neuro- fuzzy systems are adaptive networks and provide robust learning capabilities and are widely utilized in various applications such as pattern recognition, system identification, image processing and prediction. Local linear model tree (LOLIMOT) is a type of Takagi-Sugeno-Kang neuro fuzzy algorithm which has proven its efficiency compared with other neuro fuzzy networks in learning the nonlinear systems and pattern recognition. In this paper, a dedicated reconfigurable and parallel processing hardware for LOLIMOT algorithm and its applications are presented. This hardware realizes on-chip learning which gives it the capability to work as a standalone device in a system. The synthesis results on FPGA platforms show its potential to improve the speed at least 250 of times faster than software implemented algorithms.

was hosted by the National Renewable Energy Laboratory (NREL) and was attended by over 60 faculty, students and research scientists from 44 states. A record number of twelve DOE National Laboratories participated in the review to stimulate networking and partnership with these DOE crown jewels. The program was designed to achieve three major objectives: 1) Review projects currently funded by the DOE EPSCoR program, 2) Identify and stimulate interaction between the EPSCoR university community and the DOE National Laboratories, and 3) Provide a forum for exchange of information and ideas regarding the opportunities for renewable energies to contribute to the world’s future energy needs. DOE EPSCoR Program Review – A poster session was held on the afternoon and evening of Monday, July 23 rd where over 40 posters representing DOE funded EPSCoR projects were displayed. There was a lively exchange of information on a wide variety of topics ranging from photonics and electro-optic materials, magneto-inertial fusion of plasmas, coal syngas and solid oxide fuel cells, nanomagnetism, grid computing, and high-energy physics. More poster topics included cutting-edge topics on novel battery systems, wind-hydrogen energy systems, solar cells and thermochemical conversion of woody biomass to fuel, all of which were funded by DOE

1 Proposed work This was a continuation proposal to test and implement a novel multiple model framework for nonlinear plants whose dynamics are rapidly varying in a state space that is assumed known. During this research we had the opportunity to compare in the same dataset two parallel approaches to implement the gating function in the multiple model framework. We also had the opportunity to evaluate a new and intriguing nonlinear dynamical system approach called the echo state network (ESN) to implement adaptive critics. Unfortunately, we had to stop at the system identification stage, because a no cost extension request was not granted. Productivity During the time of this grant we published three journal papers and one conference proceedings, acknowledging the NASA support:

To obtain more flexibility and more effective capability of handling and processing imprecise information about the real world, fuzzy set theory is introduced into GIS. FuzzyCell is a system designed and implemented to enhance conventional GIS software (ArcMap®) with fuzzy set theory. Extending GIS with fuzzy logic (a linguistic approach as the model of human thinking) not only offers a way to represent and handle uncertainty present in the continuous real world but also assist GIS user to make decisions using experts ’ experiences in decision-making process. The cost of finding solutions to decision-making problems by models which enable decision-makers to express their constraints and imprecise concepts that are used with geographic data (i.e., fuzzy logic) for large volume is high. For such cases, artificial neural networks (ANNs), which can solve complex problems and can “learn ” from prior applications, can be used. In this study, a fuzzy rule based system (FuzzyCell) was used to model site selection problem by capturing rules from human experts. The ANNs were trained by obtained fuzzy measures against input data to recognize patterns for reproduction of relevant sites for new locations and were tested whether ANNs can produce reasonable guesses for locations other than training sites when compared to results obtained from FuzzyCell. Two metrics, Kolmogorov-Simirnov test metric and root mean squared error values, were used for testing. It was found in this study that ANNs provide reasonable guesses for locations other than training sites. 1.

2.1 Systems without feedback control.................... 8