Dimensional Representation and Gradient Boosting for Seismic Event Classification
Abstract
In this research, we conducted experiments of representational structures on 5009 seismic signals with the intent of finding a method to classify signals as either an explosion or an earthquake in an automated fashion. We also applied a gradient boosted classifier. While perfect classification was not attained (approximately 88% was our best model), some cases demonstrate that many events can be filtered out as very high probability being explosions or earthquakes, diminishing subject-matter experts'(SME) workload for first stage analysis. It is our hope that these methods can be refined, further increasing the classification probability.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2017
- Bibcode:
- 2017AGUFM.S43A0835S
- Keywords:
-
- 1914 Data mining;
- INFORMATICS;
- 1932 High-performance computing;
- INFORMATICS;
- 1942 Machine learning;
- INFORMATICS;
- 7290 Computational seismology;
- SEISMOLOGY