Clustering on dissimilarity representations for detecting mislabelled seismic signals at nevado del ruiz volcano

Abstract

Classification of seismic signals at Colombian volcanoes has been carried out manually by visual inspection. In order to reduce the workload for the seismic analysts and to turn classification reliableand objective, the use of supervised learning algorithms has been explored; particularly classifiers built in dissimilarity spaces. Nonetheless, the performance of such learning methods is subject to the availability of a representative and a priori well classified training sets. To detect mislabeled events, the use of clustering techniques on the dissimilarity representations is proposed. Our experiments,performed on re-analyzed seismic signals, show a significant improvement respect to recognition accuracies for the original data sets.