Show simple item record

dc.contributor.authorAlkhalifah, Tariq Ali
dc.contributor.authorWang, H.
dc.contributor.authorOvcharenko, Oleg
dc.date.accessioned2022-05-10T09:20:35Z
dc.date.available2022-05-10T09:20:35Z
dc.date.issued2021
dc.identifier.citationAlkhalifah, T., Wang, H., & Ovcharenko, O. (2021). MLReal: Bridging the gap between training on synthetic data and real data applications in machine learning. 82nd EAGE Annual Conference & Exhibition. https://doi.org/10.3997/2214-4609.202113262
dc.identifier.isbn9781713841449
dc.identifier.doi10.3997/2214-4609.202113262
dc.identifier.urihttp://hdl.handle.net/10754/676722
dc.description.abstractThe requirement for accurate labels in supervised learning often forces us to train our networks using synthetic data. However, synthetic experiments do not reflect the realities of the field experiment, and we end up with poor performance of the trained neural network (NN) models at the inference stage. Thus, we describe a novel approach to enhance our NN model training with real data features (domain adaptation). This is accomplished by applying two operations on the input data to the NN model, whether they are from the synthetic or real data subset class: 1) The crosscorrelation of the input data section (i.e. shot gather or seismic image) with a fixed reference trace from that section. 2) The convolution of the resulting data with a randomly chosen auto correlated section of the other subset class. In the training stage, as expected, the input data are from the synthetic subset class and the auto-corrected sections are from the real subset class, and in the inference/application stage, it is the opposite. An example application on passive seismic data for microseismic event source location determination is used to demonstrate the power of this approach in improving the applicability of our trained models on real data.
dc.description.sponsorshipWe thank Umair bin Waheed from KFUPM, Frantisek Stanek from Seismik and Claire Birnie and Yuanyuan Li from KAUST for helpful discussions. We thank Microseismic, Inc. and Newfield Exploration Mid-Continent, Inc. for graciously supplying the data. We also thank KAUST for its support.
dc.publisherEuropean Association of Geoscientists & Engineers
dc.relation.urlhttps://www.earthdoc.org/content/papers/10.3997/2214-4609.202113262
dc.rightsArchived with thanks to European Association of Geoscientists & Engineers
dc.titleMLREAL: BRIDGING THE GAP BETWEEN TRAINING ON SYNTHETIC DATA AND REAL DATA APPLICATIONS IN MACHINE LEARNING
dc.typeConference Paper
dc.contributor.departmentEarth Science and Engineering Program
dc.contributor.departmentPhysical Science and Engineering (PSE) Division
dc.conference.date2021-10-18 to 2021-10-21
dc.conference.name82nd EAGE Conference and Exhibition 2021
dc.conference.locationAmsterdam, Virtual, NLD
dc.eprint.versionPost-print
dc.identifier.volume7
dc.identifier.pages5478-5482
dc.identifier.arxivid2109.05294
kaust.personAlkhalifah, Tariq Ali
kaust.personWang, H.
kaust.personOvcharenko, Oleg
dc.identifier.eid2-s2.0-85127920485
refterms.dateFOA2022-05-23T10:22:46Z


Files in this item

Thumbnail
Name:
eageabs.pdf
Size:
3.581Mb
Format:
PDF
Description:
Accepted Manuscript

This item appears in the following Collection(s)

Show simple item record