Robert Orchard

Exploring medical data using visual spaces with genetic programming and implicit functional mappings

Two medical data sets (Breast cancer and Colon cancer) are investigated within a visual data mining paradigm through the unsupervised construction of virtual reality spaces using genetic programming and classical optimization (for comparison purposes). The desired visual spaces are such that a modified genetic programming approach was proposed in order to generate programs representing vector functions. The extension leads to populations that are composed of forests, instead of single expression trees. No particular kind of genetic programming algorithm is required due to the generic nature of the approach taken in the paper. The results (visual spaces) show that the relationships between the data objects and their classes can be appreciated in all of the obtained spaces regardless of the mapping error. In addition, the spaces obtained with genetic programming resulted in lower mapping errors than a classical optimizer and produced relatively simple equations. Further, the set of obtained equations can be statistically analyzed in terms of the original attributes in order to further the understanding of the derivation of the new nonlinear features that are constructed. Thus, explicit mappings provided by genetic programming can be used for feature selection and generation in data mining where scalar and/or vector functions are involved.

Virtual reality high dimensional objective spaces for multi-objective optimization: An improved representation

This paper presents an approach for constructing improved visual representations of high dimensional objective spaces using virtual reality. These spaces arise from the solution of multi-objective optimization problems with more than 3 objective functions which lead to high dimensional Pareto fronts. The 3-D representations of m-dimensional Pareto fronts, or their approximations, are constructed via similarity structure mappings between the original objective spaces and the 3-D space. Alpha shapes are introduced for the representation and compared with previous approaches based on convex hulls. In addition, the mappings minimizing a measure of the amount of dissimilarity loss are obtained via genetic programming. This approach is preliminarily investigated using both theoretically derived high dimensional Pareto fronts for a test problem (DTLZ2) and practically obtained objective spaces for the 4 dimensional knapsack problem via multi-objective evolutionary algorithms like HLGA, NSGA, and VEGA. The improved representation captures more accurately the real nature of the m-dimensional objective spaces and the quality of the mappings obtained with genetic programming is equivalent to those computed with classical optimization algorithms.

Automated case base creation and management

In this paper, we report on a scheme for automated case base creation and management. The scheme aims at reducing the difficulty and human effort required for case creation. This paper provides an overview of the proposed scheme and outlines its technical implementation as an automated case creation system for the Integrated Diagnostic System. Some experimental results for testing the scheme and an interactive tool for evaluating the constructed case base are presented.

2009 Special Issue: Neural networks with multiple general neuron models: A hybrid computational intelligence approach using Genetic Programming

Classical neural networks are composed of neurons whose nature is determined by a certain function (the neuron model), usually pre-specified. In this paper, a type of neural network (NN-GP) is presented in which: (i) each neuron may have its own neuron model in the form of a general function, (ii) any layout (i.e network interconnection) is possible, and (iii) no bias nodes or weights are associated to the connections, neurons or layers. The general functions associated to a neuron are learned by searching a function space. They are not provided a priori, but are rather built as part of an Evolutionary Computation process based on Genetic Programming. The resulting network solutions are evaluated based on a fitness measure, which may, for example, be based on classification or regression errors. Two real-world examples are presented to illustrate the promising behaviour on classification problems via construction of a low-dimensional representation of a high-dimensional parameter space associated to the set of all network solutions.

Authoring cases from free-text maintenance data

Automatically authoring or acquiring cases in the case-based reasoning (CBR) systems is recognized as a bottleneck issue that can determine whether a CBR system will be successful or not. In order to reduce human effort required for authoring the cases, we propose a framework for authoring the case from the unstructured, free-text, historic maintenance data by applying natural language processing technology. This paper provides an overview of the proposed framework, and outlines its implementation, an automated case creation system for the Integrated Diagnostic System. Some experimental results for testing the framework are also presented.

Characterization of climatic variations in Spain at the regional scale: A computational intelligence approach

Computational intelligence and other data mining techniques are used for characterizing regional and time-varying climatic variations in Spain in the period 1901-2005. Daily maximum temperature data from 10 climatic stations are analyzed (with and without missing values) using principal components (PC), similarity-preservation feature generation, clustering, Kolmogorov-Smirnov dissimilarity analysis and genetic programming (GP). The new features were computed using hybrid optimization (differential evolution and Fletcher-Reeves) and GP. From them, a scalar regional climatic index was obtained which identifies time landmarks and changes in the climate rhythm. The equations obtained with GP are simpler than those obtained with PC and they highlight the most important sites characterizing the regional climate. Whereas the general consensus is that there has been a clear and smooth trend towards warming during the last decades, the results suggest that the picture may probably be much more complicated than what is usually assumed.

Learning the neuron functions within a neural network via Genetic Programming: Applications to geophysics and hydrogeology

A neural network classifier is sought. Classical neural network neurons are aggregations of a weight multiplied by an input value and then controlled via an activation function. This paper learns everything within the neuron using a variant of Genetic Programming called Gene Expression Programming. That is, this paper does not explicitly use weights or activation functions within a neuron, nor bias nodes within a layer. Promising preliminary results are reported for a study of the detection of underground caves (a 1 class problem) and for a study of the interaction of water and minerals near a glacier in the Arctic (a 5 class problem).