Automatic classification of seismic events by neural networks
Abstract
The automatic classification of seismic events recorded in an active volcanic area is investigated by means of neural networks. An approach based on multilayer perceptrons is shown. Typical waveforms of seismic events are illustrated. Events whose time series could be correctly classified by seismologists become ambiguous when only the corresponding autocorrelation function is taken into account. Based on this consideration an improvement of the percentage of success can be achieved by integrating the pattern obtained considering the autocorrelation function of the shocks with information derived from the shape of the signals.
- Publication:
-
IGARSS 1992; Proceedings of the 12th Annual International Geoscience and Remote Sensing Symposium
- Pub Date:
- 1992
- Bibcode:
- 1992igrs.conf..224F
- Keywords:
-
- Neural Nets;
- Seismology;
- Self Organizing Systems;
- Volcanoes;
- Autocorrelation;
- Classifying;
- Machine Learning;
- Signal Processing;
- Volcanology