Conventional Feature Selection Methods Research Articles

A new genetic feature selection with neural network ensemble

A neural network ensemble is a learning paradigm in which a finite collection of neural networks is trained for the same task. Ensembles generally show better classification and generalization performance than a single neural network does. In this paper, a new feature selection method for a neural network ensemble is proposed for pattern classification. The proposed method selects an adequate feature subset for each constituent neural network of the ensemble using a genetic algorithm. Unlike the conventional feature selection method, each neural network is only allowed to have some (not all) of the considered features. The proposed method can therefore be applied to huge-scale feature classification problems. Experiments are performed with four databases to illustrate the performance of the proposed method.

Linear and nonlinear quantitative structure–property relationship models for solubility of some anthraquinone, anthrone and xanthone derivatives in supercritical carbon dioxide

A quantitative structure–property relation (QSPR) study was conducted on the solubility in supercritical fluid carbon dioxide (SCF-CO 2) of some recently synthesized anthraquinone, anthrone and xanthone derivatives. The data set consisted of 29 molecules in various temperatures and pressures, which form 1190 solubility data. The combined data splitting-feature selection (CDFS) strategy, which previously developed in our research group, was used as descriptor selection and model development method. Modeling of the relationship between selected molecular descriptors and solubility data was achieved by linear (multiple linear regression; MLR) and nonlinear (artificial neural network; ANN) methods. The QSPR models were validated by cross-validation as well as application of the models to predict the solubility of three external set compounds, which did not have contribution in model development steps. Both linear and nonlinear methods resulted in accurate prediction whereas more accurate results were obtained by ANN model. The respective root mean square error of prediction obtained by MLR and ANN models were 0.284 and 0.095 in the term of logarithm of g solute m −3 of SCF-CO 2. A comparison was made between the models selected by CDFS method and the conventional stepwise feature selection method. It was found that the latter produced models with higher number of descriptors and lowered prediction ability, thus it can be considered as an over-fitted model.

Efficient huge-scale feature selection with speciated genetic algorithm

With increasing interest in bioinformatics, sophisticated tools are required to efficiently analyze gene information. The classification of gene expression profiles is crucial in those fields. Since the features of data obtained by microarray technology come to be over thousands, it is essential to extract useful information by selecting proper features. The information without any feature selection might be redundant so that this can deteriorate the performance of classification. The conventional feature selection method with genetic algorithm has difficulty for huge-scale feature selection. In this paper, we modify the representation of chromosome to be suitable for huge-scale feature selection and adopt speciation to enhance the performance of feature selection by obtaining diverse solutions. Experimental results with DNA microarray data from cancer patients show that the selected genes by the proposed method are useful for cancer classification.

Using neural networks to select wavelet features for breast cancer diagnosis

This study focuses on improving microcalcification classification by establishing an efficient computer-aided diagnosis system that extracts Daubechies-4 and biorthogonal wavelet features. These wavelets were chosen because they have been used in military target recognition and fingerprint recognition research with images characterized by low contrast, similar to mammography. Feature selection techniques are employed to further increase classification performance. The artificial neural network feature selection techniques are complemented by a conventional decision boundary-based feature selection method. The results using the wavelet features are compared to more conventional measures of image texture, angular second moment, and Karhunen Loeve coefficients. The use of alternative signal processing to compare wavelet and neural techniques allows for a measure of the problem difficulty. It is concluded that advances and contributions have been made with the introduction of two novel feature extraction methods for breast cancer diagnosis, wavelets and eigenmasses. Additionally, feature selection techniques are demonstrated, compared, and validated, transforming adequate discrimination power into promising classification results.