The paper presents a novel approach to feature extraction using genetic algorithms (GAs) that simultaneously performs feature selection, feature extraction, and classifier training. The GA optimizes a vector of feature weights, which are used to scale the original features in either linear or nonlinear fashion. A masking vector is also employed to select a subset of features. The technique is evaluated in combination with the k-nearest neighbor (kNN) classification rule and compared with classical feature selection and extraction techniques, including sequential floating forward feature selection and linear discriminant analysis. The effectiveness of the GA-based method is demonstrated through experiments on medical datasets and a challenging biochemistry problem involving the identification of favorable water-binding sites on protein surfaces. The results show that the GA-kNN approach achieves high classification accuracy while requiring fewer features compared to other methods, making it both accurate and efficient. The paper also discusses the advantages of the GA feature extraction technique, such as the explicit relationships between original and transformed features, and suggests potential extensions, including the use of other classification techniques and direct representation of the transformation matrix.The paper presents a novel approach to feature extraction using genetic algorithms (GAs) that simultaneously performs feature selection, feature extraction, and classifier training. The GA optimizes a vector of feature weights, which are used to scale the original features in either linear or nonlinear fashion. A masking vector is also employed to select a subset of features. The technique is evaluated in combination with the k-nearest neighbor (kNN) classification rule and compared with classical feature selection and extraction techniques, including sequential floating forward feature selection and linear discriminant analysis. The effectiveness of the GA-based method is demonstrated through experiments on medical datasets and a challenging biochemistry problem involving the identification of favorable water-binding sites on protein surfaces. The results show that the GA-kNN approach achieves high classification accuracy while requiring fewer features compared to other methods, making it both accurate and efficient. The paper also discusses the advantages of the GA feature extraction technique, such as the explicit relationships between original and transformed features, and suggests potential extensions, including the use of other classification techniques and direct representation of the transformation matrix.