The generation of a set of rules underlying a classification problem is performed by applying a new algorithm, called Hamming Clustering (HC). It reconstructs the and-or expression associated with any Boolean function from a training set of samples. The basic

Neural networks have been successfully applied to solve a variety of application problems including classification and function approximation. They are especially useful as function approximators because they do not require prior knowledge of the input data distribution and they have been shown to be universal approximators. In many applications, it is desirable to extract knowledge that can explain how the problems are solved by the networks. Most existing approaches have focused on extracting symbolic rules for classification. Few methods have been devised to extract rules from trained neural networks for regression. This article presents an approach for extracting rules from trained neural networks for regression. Each rule in the extracted rule set corresponds to a subregion of the input space and a linear function involving the relevant input attributes of the data approximates the network output for all data samples in this subregion. Extensive experimental results on 32 benchmark data sets demonstrate the effectiveness of the proposed approach in generating accurate regression rules.

A neuro-fuzzy methodology is described which involves connectionist minimization of a fuzzy feature evaluation index with unsupervised training. The concept of a flexible membership function incorporating weighed distance is introduced in the evaluation index to make the modeling of clusters more appropriate. A set of optimal weighing coefficients in terms of networks parameters representing individual feature importance is obtained through connectionist minimization. Besides, the investigation includes the development of another algorithm for ranking of different feature subsets using the aforesaid fuzzy evaluation index without neural networks. Results demonstrating the effectiveness of the algorithms for various real life data are provided

This paper describes two examples of real-life applications of texture segmentation using M-band wavelets. In the first part of the paper, an efficient and computationally fast method for segmenting text and graphics part of a document image based on textural cues is presented. It is logical to assume that the graphics part has different textural properties than the non-graphics (text) part. So, this is basically a two-class texture segmentation problem. The second part of the paper describes a segmentation scheme for another real-life data such as remotely sensed image. Different quasi-homogeneous regions in the image can be treated to have different texture properties. Based on this assumption the multi-class texture segmentation scheme is applied for this purpose. Keywords: Texture segmentation; M-band wavelets; document image; remotely sensed image. AMS Subject Classification: 22E46, 53C35, 57S20 1.

In this article we introduce partial retraining, an algorithm to determine the relevance of the input variables of a trained neural network. We place this algorithm in the context of other approaches to relevance determination. Numerical experiments on both artificial and real-world problems show that partial retraining outperforms its competitors, which include methods based on constant substitution, analysis of weight magnitudes, and "optimal brain surgeon". 1 Introduction Feedforward neural networks are able to learn the relationship between input and output variables. Even when knowledge about the problem is limited, as for example in cases where no explicit physical or economical model can be built, neural networks may still capture some of the underlying principles. Especially with a lack of domain knowledge, the usual approach in neural network modeling is to include all input variables that may have an effect on the output. This approach is suboptimal in several aspects. First...

The extraction of a set of rules underlying a classification problem is performed by applying a new algorithm reconstructing the and-or expression of any Boolean function from a given set of samples.

Abstract Neural networks have been successfully applied to solve a variety of application problems including classification and function approximation. They are especially useful as function approximators because they do not require prior knowledge of the input data distribution and they have been shown to be universal approximators. In many applications, it is desirable to extract knowledge that can explain how the problems are solved by the networks. Most existing approaches have focused on extracting symbolic rules for classification. Few methods have been devised to extract rules from trained neural networks for regression. This article presents an approach for extracting rules from trained neural networks for regression. Each rule in the extracted rule set corresponds to a subregion of the input space and a linear function involving the relevant input attributes of the data approximates the network output for all data samples in this subregion. Extensive experimental results on 32 benchmark data sets demonstrate the effectiveness of the proposed approach in generating accurate regression rules. Index terms: Regression, network pruning, rule extraction. \Lambda This author's work was sponsored in part by the Department of the Navy, Office of Naval Research, Grant N00014-98-1-0568. The content of this information does not necessarily reflect the position of the Government.

A new algorithm, called Hamming Clustering (HC), is proposed to extract a set of rules underlying a given classification problem. It is able to reconstruct the and-or expression associated with any Boolean function from a training set of samples.

Abstract In many problems from different disciplines such as engineering, physics, medicine, and biology, a series of experimental data is used in order to generate a model that can describe a system with minimum noise. The procedure for building a model provides a description of the behavior of the system under study and can be used to give a prediction for the future. Herein a novel hierarchical bi-level implementation of the cross validation method is presented. In this bi-level schema, the leader optimization problem builds (training) the model and the follower checks (testing) the developed model. The problem of synthesis and analysis of regulatory networks is used to compare the classical cross validation method to the proposed methodology referred to as bi-level cross validation. In all the examples considered, the bi-level cross validation results in a better model compared with the classical cross validation approach.