A one class KNN for signal identification: a biological case study

The paper describes an application of a one class KNN to identifydifferent signal patterns embedded in a noise structured background. Theproblem becomes harder whenever only one pattern is well-represented in thesignal; in such cases, one class classifier techniques are more indicated. Theclassification phase is applied after a preprocessing phase based on a multilayer model (MLM) that provides preliminary signal segmentation in an intervalfeature space. The one class KNN has been tested on synthetic and real(Saccharomyces cerevisiae) microarray data in the specific problem of DNAnucleosome and linker regions identification. Results have shown, in bothcases, a good recognition rate.