Resampling methods for quality assessment of classifier performance and optimal number of features

• Published in: Signal processing Vol. 93; no. 11; pp. 2956 - 2968
• Main Authors: Fandos, Raquel, Debes, Christian, Zoubir, Abdelhak M.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.11.2013; Elsevier
• Subjects: Algorithms; Applied sciences; Bootstrap; Classification; Classifier design and evaluation; Classifiers; Confidence intervals; Detection, estimation, filtering, equalization, prediction; Exact sciences and technology; Feature evaluation and selection; Information, signal and communications theory; Optimal dimensionality; Optimization; Pattern recognition; Quality assessment; Resampling; Signal and communications theory; Signal processing; Signal representation. Spectral analysis; Signal, noise; Telecommunications and information theory; Bootstrap; Feature evaluation and selection; Optimal dimensionality; Pattern recognition; Resampling; Classifier design and evaluation; Performance evaluation; Automatic classification; Dimensionality; Probabilistic approach; Probability distribution; Algorithm; Signal classification; Confidence interval; Bootstrapping; Quality control; Testing equipment; Database; Signal processing; Feature extraction; Resampling method; Figure of merit

We address two fundamental design issues of a classification system: the choice of the classifier and the dimensionality of the optimal feature subset. Resampling techniques are applied to estimate both the probability distribution of the misclassification rate (or any other figure of merit of a cla...