Data Science In High Energy Physics (INVITED)

Abstract

Dr. Hakl will consider a special neural network classifier with switching units and demonstrate application of this model to signal separation of high speed triggering of events in high energy physics in a D0-experiment at Fermilabs in Chicago. Dr. Coufal will consider particle filtering with radial fuzzy systems and kernel density estimates, a family of tractable systems due to their shape preservation property. Kernel density estimates of the filtering density builds on Fourier analysis to obtain estimates of their theoretical counterparts despite the non-i.i.d. character of the samples. Dr. Jirina will address properties of distance-based classifiers on fractal data with a weighted k-NN rule classifier, based on inherent transformation in multidimensional space of general data to a uniform multidimensional space, where it is easier to classify data. Some applications to particle physics will be shown.