A SEISMIC CLASSIFICATION MODEL

The report is intended as an introduction to one possible approach to the seismic classification problem. It develops a very general classification model using automatic non-parametric learning based on limited data of known classification. The model accepts discriminants extracted from the seismogram and yields the probability that the input was due to an earthquake or an explosion. Thus, the discriminants are assumed to be available as inputs. Pattern recognition as used here is defined, the classification procedure is outlined, the adaptive estimation of joint probability-densities from a finite number of multi-dimensional vectors of known classification the learning model is discussed, a simplified flow diagram of the learning model is presented, and the selection of necessary control parameters is investigated.