A Novel Approach to Design Classifiers Using Genetic Programming

Abstract

Abstract-We propose a new approach for designing classifiers for a -class ( 2) problem using genetic programming (GP). The proposed approach takes an integrated view of all classes when the GP evolves. A multitree representation of chromosomes is used. In this context, we propose a modified crossover operation and a new mutation operation that reduces the destructive nature of conventional genetic operations. We use a new concept of unfitness of a tree to select trees for genetic operations. This gives more opportunity to unfit trees to become fit. A new concept of OR-ing chromosomes in the terminal population is introduced, which enables us to get a classifier with better performance. Finally, a weight-based scheme and some heuristic rules characterizing typical ambiguous situations are used for conflict resolution. The classifier is capable of saying "don't know" when faced with unfamiliar examples. The effectiveness of our scheme is demonstrated on several real data sets.

Looking for Prototypes by Genetic Programming

“…The proposed approach has been compared with another GP based approach previously proposed in [10]. Furthermore, the results obtained by the preliminary version of the method [12] are also shown for comparison.…”

“…Hence, 100 runs have been performed for each data set. Note that, in [10], the number of prototypes is a priori fixed, while in our method it is automatically found. The results show that the proposed method outperforms those used for comparison on all the data sets taken into account, confirming the validity of the approach.…”

“…Only recently, classification problems have been faced by using GP. In [10], GP has been used to evolve equations (encoded as derivation trees) involving simple arithmetic operators and feature variables. The method was tested on different type of data, including images.…”

“…The method presented here has been tested on three publicly available databases and the classification results have been compared with those obtained by the preliminary version of the method and with another GP based method presented in the literature [10].…”

Automatic Generation of Cognitive Theories using Genetic Programming

“…As evolutionary computation is not as sensitive to local minima and initial conditions as other hill-climbing methods (Koza, 1992), and as it can explore large search spaces efficiently and in parallel, it is ideal in problems where the information is noisy and subject to uncertainty. Evolutionary computation in general and GP in particular have already been used for a wide variety of applications including digital hardware design and optimization (Jackson, 2005), analog hardware design and optimization (Dastidar et al, 2005), solving multiobjective problems (Whigham and Crapper, 2001), design of classifiers (Muni et al, 2004), and also some neuroscientific applications like diagnostic discovery (Kentala et al, 1999), neuromuscular disorders assessment (Pattichis and Schizas, 1996), and interpretation of magnetic-resonance brain images (Sonka et al 1996).…”