Authors:
Claus Aranha
1
;
Yuri Cossich Lavinas
2
;
Marcelo Ladeira
2
and
Bogdan Enescu
1
Affiliations:
1
University of Tsukuba, Japan
;
2
University of Brasilia, Brazil
Keyword(s):
Earthquakes, Forecast Model, Genetic Algorithm, Application.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Computational Intelligence
;
Evolutionary Computing
;
Genetic Algorithms
;
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Knowledge Discovery and Information Retrieval
;
Knowledge-Based Systems
;
Machine Learning
;
Soft Computing
;
Symbolic Systems
Abstract:
Understanding the mechanisms and patterns of earthquake occurrence is of crucial importance for assessing and mitigating the seismic risk. In this work we analyze the viability of using Evolutionary Computation (EC) as a means of generating models for the occurrence of earthquakes. Our proposal is made in the context of the "Collaboratory for the Study of Earthquake Predictability" (CSEP), an international effort to standardize the study and testing of earthquake forecasting models. We use a standard Genetic Algorithm (GA) with real valued genome, where each allele corresponds to a bin in the forecast model. The design of an appropriate fitness function is the main challenge for this task, and we describe two different proposals based on the log-likelihood of the candidate model against the training data set. The resulting forecasts are compared with the Relative Intensity algorithm, which is traditionally employed by the CSEP community as a benchmark, using data from the Japan Meteo
rological Agency (JMA) earthquake catalog. The forecasts generated by the GA were competitive with the benchmarks, specially in scenarios with a large amount of inland seismic activity.
(More)