Training a Φ-machine classifier using feature scaling-space

Efficient classification of signal patterns plays a vital role in data mining and other computational intelligence applications. This paper presents a reciprocal-sigmoid model for pattern classification. The proposed classifier can be considered as a «-machine since it preserves the theoretical advantage of linear machines where the weight parameters can be estimated in a single step. To handle possible over-fitting when using high order models, the classifier is trained using multiple samples of uniformly scaled pattern features. The classifier is empirically evaluated using benchmark data sets for statistical evidence.