Outlier detection from a mixture distribution when training data are unlabeled

We consider the difficult task of using seismic signals (or any other discr iminants) for detecting nuclear explosions from the large number of backgro und signals such as earthquakes and mining blasts. Given a ground-truth dat abase (i.e., labeled data), Fisk et al. (1996) consider the problem of dete cting outliers (nuclear explosions) from a single background-signal populat ion, and their approach has been applied successfully in several. regions a round the world. Wang et al. (1997) attack the problem in terms of modeling the background as a mixture distribution and looking for outliers (nuclear events) from that: mixture, However, those authors only considered the cas e in which at least some Traction of the training sample was labeled, that is, at least some ground-truth information was available, and the number of distinct classes of events was known. In the current article, we extend th ese results to the case in which no events in the training sample are label ed and also to the ease in which the number of event types represented in t he training sample is unknown. one can view the mixture approach as a robus t method for fitting a density to training data that may: not be normally d istributed whether or not the data consist of identifiable components that have a physical interpretation. The technique is demonstrated using simulat ed data as well as two sets of seismic data.