Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/ UPCommons. Portal de...arrow_drop_down
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
versions View all 4 versions
addClaim

Deep Learning for earthquake detection

Authors: Mus León, Sergi;

Deep Learning for earthquake detection

Abstract

Models realistes per la detecció automàtica de sismes només es poden aconseguir analitzant un gran volum de dades. Aquestes dades són difícils de processar amb un catalogat manual. Aquest projecte proposa la automatització d'aquest procés mitjançant l'ús de dades reals i tècniques de DL. Estudis anteriors demostren que els millors resultats per resoldre problemes d'aquest tipus s'obtenen utilitzant CNN. A part d'aquestes xarxes, el projecte implementa unes xarxes més simples, com les FFNN, esperant obtenir resultats competitius en comparació als obtinguts per les CNN. Per obtenir aquest objectiu, el projecte proposa una tècnica de preprocessament de traces sísmiques innovadora. Els resultats mostren que les CNN arriben a detectar el 98,2% dels events, mentre que les FFNN ho fan en un 91,2%. Per un altra part, el projecte inclou un algoritme basat en codificadors per aproximar el temps de arribada del event a la estació. Finalment, el projecte utilitza la metodologia de meta-learning per realitzar la detecció de sismes per les dades que provenen exclusivament d'una única estació.

Modelos realistas para la detección automática de sismos solo se pueden conseguir anal-izando grandes cantidades de datos. Estos datos son difíciles de procesar por catalogado manual. Este proyecto plantea la automatización de este proceso usando datos reales y técnicas de DL. Estudios anteriores demuestran que los mejores resultados, para resolver este tipo de problemas, se alcanzan usando CNN. A parte de estas redes, el proyecto implementa redes más simples, como las FFNN, esperando obtener resultados competitivos respecto a los obtenidos por las CNN. Para alcanzar este objetivo, el proyecto propone una técnica de preprocesamiento novedosa de trazas sísmicas. Los resultados muestran que CNN llega a detectar el 98,2% de los eventos, mientras que FFNN lo realiza en un 91,2%. Por otra parte, el proyecto incluye un algoritmo basado en codificadores para aproximar el tiempo de arrivo del evento a la estación. Finalmente, el proyecto utiliza la metodología de meta-learning para realizar detección de sismos para los datos que provienen de una única estación.

The creation of accurate ground motion models can only be achieved with the help ofvast amounts of labelled data. The manual cataloging makes the processing of this data.The project proposes the automation of the labeling through real data and DL.Previous study show that CNN are the best architecture for this type of problems. Theproject also implements FFNN, a simpler architecture, with the intent of achieving com-petitive results. To make this objective possible the project proposes a novel preprocessing.The results show that CNN reach an accuracy of 98.2%, while FFNN achieves 91.2%.Moreover, the project includes an encoder based algorithm to approximate arrival timesto the station. Finally, the project make use of meta-learning to detect seismic eventsproviding from a single station.

Country
Spain
Keywords

Feed Forward Neural Network, :Enginyeria de la telecomunicació::Telemàtica i xarxes d'ordinadors [Àrees temàtiques de la UPC], Programació (Ordinadors), Deep learning, Convolutional Neural Network, Computer programming, Neural networks (Computer science), Àrees temàtiques de la UPC::Enginyeria de la telecomunicació::Telemàtica i xarxes d'ordinadors, Deep Learning, Àrees temàtiques de la UPC::Enginyeria civil::Geotècnia::Sismologia, Xarxes neuronals (Informàtica), :Enginyeria civil::Geotècnia::Sismologia [Àrees temàtiques de la UPC], Sismologia, Seismology, Aprenentatge profund

  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
    OpenAIRE UsageCounts
    Usage byUsageCounts
    visibility views 49
    download downloads 59
  • 49
    views
    59
    downloads
    Powered byOpenAIRE UsageCounts
Powered by OpenAIRE graph
Found an issue? Give us feedback
visibility
download
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
views
OpenAIRE UsageCountsViews provided by UsageCounts
downloads
OpenAIRE UsageCountsDownloads provided by UsageCounts
0
Average
Average
Average
49
59
Green