Face recognition through a chaotic neural network model

K-sets models are connectionist methods based on neuron populations, conceived through EEG analyses of the mammalian olfactory system. These models present a biologically more plausible structure and behavior than other neural networks models. K-sets have been used in many machine-learning problems, with potential application on pattern recognition while presenting novel chaotic concepts for signal processing. This paper presents the characteristics of the K-sets models and their application in a face recognition task. Our method was tested using Yale Face Database B and the results show that it outperforms popular recognition methods.