Unsupervised learning models-based CRM anomaly detection using GPU

Deep learning models have improved several business intelligence tools like Customer relationship Management(CRM) systems. However, those models have increased the need for advanced computational capacity and infrastructure. Modern accelerators are starting to have floating-point precision arithmeti...

Full description

Autores:
Tipo de recurso:
Article of investigation
Fecha de publicación:
2021
Institución:
Universidad de Bogotá Jorge Tadeo Lozano
Repositorio:
Expeditio: repositorio UTadeo
Idioma:
eng
OAI Identifier:
oai:expeditiorepositorio.utadeo.edu.co:20.500.12010/20764
Acceso en línea:
http://hdl.handle.net/20.500.12010/20764
http://expeditio.utadeo.edu.co
Palabra clave:
Analítica de datos
Análisis de sistemas -- Tesis y disertaciones académicas
Teoría de sistemas -- Tesis y disertaciones académicas
Diseño de sistemas -- Tesis y disertaciones académicas
Rights
License
Abierto (Texto Completo)
id UTADEO2_ece4fe34befa17f259171324234d752e
oai_identifier_str oai:expeditiorepositorio.utadeo.edu.co:20.500.12010/20764
network_acronym_str UTADEO2
network_name_str Expeditio: repositorio UTadeo
repository_id_str
dc.title.spa.fl_str_mv Unsupervised learning models-based CRM anomaly detection using GPU
title Unsupervised learning models-based CRM anomaly detection using GPU
spellingShingle Unsupervised learning models-based CRM anomaly detection using GPU
Analítica de datos
Análisis de sistemas -- Tesis y disertaciones académicas
Teoría de sistemas -- Tesis y disertaciones académicas
Diseño de sistemas -- Tesis y disertaciones académicas
title_short Unsupervised learning models-based CRM anomaly detection using GPU
title_full Unsupervised learning models-based CRM anomaly detection using GPU
title_fullStr Unsupervised learning models-based CRM anomaly detection using GPU
title_full_unstemmed Unsupervised learning models-based CRM anomaly detection using GPU
title_sort Unsupervised learning models-based CRM anomaly detection using GPU
dc.subject.spa.fl_str_mv Analítica de datos
topic Analítica de datos
Análisis de sistemas -- Tesis y disertaciones académicas
Teoría de sistemas -- Tesis y disertaciones académicas
Diseño de sistemas -- Tesis y disertaciones académicas
dc.subject.lemb.spa.fl_str_mv Análisis de sistemas -- Tesis y disertaciones académicas
Teoría de sistemas -- Tesis y disertaciones académicas
Diseño de sistemas -- Tesis y disertaciones académicas
description Deep learning models have improved several business intelligence tools like Customer relationship Management(CRM) systems. However, those models have increased the need for advanced computational capacity and infrastructure. Modern accelerators are starting to have floating-point precision arithmetic problems generated by highly streamlined systems, powered by the need to process an ever-increasing volume of data and increasingly complex models to attend to the necessity to identify customer data that allow consolidating products or services. We focus on CRM anomalies detection using GPU(Graphics Processor Unit) because they are a relevant source of money drain for organizations and directly affect the relationship between clients and suppliers. Our results present the combination of deep learning models with a computational structure that could access by organizations, but with a combination that reduces the number of features that achieve answers to CRM system.
publishDate 2021
dc.date.accessioned.none.fl_str_mv 2021-07-29T15:33:03Z
dc.date.available.none.fl_str_mv 2021-07-29T15:33:03Z
dc.date.created.none.fl_str_mv 2021
dc.type.local.spa.fl_str_mv Trabajo de grado de maestría
dc.type.driver.spa.fl_str_mv info:eu-repo/semantics/masterThesis
dc.type.coar.spa.fl_str_mv http://purl.org/coar/resource_type/c_2df8fbb1
format http://purl.org/coar/resource_type/c_2df8fbb1
dc.identifier.uri.none.fl_str_mv http://hdl.handle.net/20.500.12010/20764
dc.identifier.repourl.spa.fl_str_mv http://expeditio.utadeo.edu.co
url http://hdl.handle.net/20.500.12010/20764
http://expeditio.utadeo.edu.co
dc.language.iso.spa.fl_str_mv eng
language eng
dc.relation.references.spa.fl_str_mv Akhtar, M.S., Chauhan, D.S., Ekbal, A., 2020. A deep multi-task contextual attention framework for multi-modal affect analysis. ACM Trans. Knowl. Discov. Data 14. doi:10.1145/3380744.
Navarro, C.A., Carrasco, R., Barrientos, R.J., Riquelme, J.A., Vega, R., 2021. Gpu tensor cores for fast arithmetic reductions. IEEE Transactions on Parallel and Distributed Systems 32, 72–84. doi:10.1109/ TPDS.2020.3011893.
Nevin, J., 1995. Relationship marketing and distribution channels: Exploring fundamental issues. Journal of the Academy of Marketing Science 23, 327–334. doi:10.1177/009207039502300413.
Ordóñez, F., Roggen, D., 2016. Deep convolutional and lstm recurrent neural networks for multimodal wearable activity recognition. Sensors (Switzerland) 16. doi:10.3390/s16010115.
Ozar, B., . Sql-server-first-responder-kit. https://github.com/ BrentOzarULTD/SQL-Server-First-Responder-Kit/blob/dev/sp_ ineachdb.sql.
Pang, G., Shen, C., Cao, L., Hengel, A.V.D., 2021. Deep learning for anomaly detection: A review. ACM Comput. Surv. 54. doi:10.1145/ 3439950.
Payne, A., Frow, P., 2005. A strategic framework for customer relationship management. Journal of Marketing 69, 167–176. doi:10. 1509/jmkg.2005.69.4.167.
Racherla, P., Hu, C., 2008. ecrm system adoption by hospitality organizations: A technology-organization-environment (toe) framework. Journal of Hospitality&Leisure Marketing 17, 30–58. URL: https:// doi.org/10.1080/10507050801978372, doi:10.1080/10507050801978372.
Raina, R., Madhavan, A., Ng, A., 2009. Large-scale deep unsupervised learning using graphics processors, in: Proceedings of the 26th International Conference On Machine Learning, ICML 2009, p. 110. doi:10.1145/1553374.1553486.
Sadhu, V., Misu, T., Pompili, D., 2019. Deep multi-task learning for anomalous driving detection using can bus scalar sensor data, in: 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 2038–2043. doi:10.1109/IROS40897.2019. 8967753.
Sagheer, A., Kotb, M., 2019. Unsupervised pre-training of a deep lstm-based stacked autoencoder for multivariate time series forecasting problems. Scientific Reports 9, 19038. URL: https://doi.org/ 10.1038/s41598-019-55320-6, doi:10.1038/s41598-019-55320-6.
Trainor, K., Andzulis, J., Rapp, A., Agnihotri, R., 2014. Social media technology usage and customer relationship performance: A capabilities-based examination of social crm. Journal of Business Research 67, 1201–1208. doi:10.1016/j.jbusres.2013.05.002.
Wang, R., Nie, K., Wang, T., Yang, Y., Long, B., 2020. Deep learning for anomaly detection, in: Proceedings of the 13th International Conference on Web Search and Data Mining, p. 894–896. doi:10.1145/3336191.3371876.
Wirth, R., Hipp, J., 2000. Crisp-dm: Towards a standard process model for data mining, in: Proceedings of the 4th international conference on the practical applications of knowledge discovery and data mining, Springer-Verlag London, UK. pp. 29–39.
Zachariadis, O., Satpute, N., Gómez-Luna, J., Olivares, J., 2020. Accelerating sparse matrix–matrix multiplication with gpu tensor cores. Computers & Electrical Engineering 88, 106848. doi:https://doi. org/10.1016/j.compeleceng.2020.106848.
Zhang, S., Wu, P., 2019. High accuracy low precision qr factorization and least square solver on gpu with tensorcore. ArXiv abs/1912.05508.
Zheng, Z., Pu, J., Liu, L., Wang, D., Mei, X., Zhang, S., Dai, Q., 2020. Contextual anomaly detection in solder paste inspection with multi-task learning. ACM Trans. Intell. Syst. Technol. 11. URL: https://doi.org/10.1145/3383261, doi:10.1145/3383261.
Brereton, R.G., Lloyd, G.R., 2016. Re-evaluating the role of the mahalanobis distance measure. Journal of Chemometrics 30, 134–143. doi:https://doi.org/10.1002/cem.2779.
Chen, I., Popovich, K., 2003. Understanding customer relationship management (crm): People, process and technology. Business Process Management Journal 9, 672–688. doi:10.1108/ 14637150310496758.
Chen, Z.Y., Fan, Z.P., Sun, M., 2012. A hierarchical multiple kernel support vector machine for customer churn prediction using longitudinal behavioral data. European Journal of Operational Research 223, 461–472. doi:https://doi.org/10.1016/j.ejor.2012.06.040.
dc.rights.coar.fl_str_mv http://purl.org/coar/access_right/c_abf2
dc.rights.local.spa.fl_str_mv Abierto (Texto Completo)
rights_invalid_str_mv Abierto (Texto Completo)
http://purl.org/coar/access_right/c_abf2
dc.format.extent.spa.fl_str_mv 12 páginas
dc.format.mimetype.spa.fl_str_mv application/pdf
dc.coverage.spatial.spa.fl_str_mv Colombia
dc.publisher.spa.fl_str_mv Universidad de Bogotá Jorge Tadeo Lozano
dc.publisher.program.spa.fl_str_mv Maestría en Ingeniería y Analítica de Datos
dc.source.spa.fl_str_mv instname:Universidad de Bogotá Jorge Tadeo Lozano
reponame:Expeditio Repositorio Institucional UJTL
instname_str Universidad de Bogotá Jorge Tadeo Lozano
institution Universidad de Bogotá Jorge Tadeo Lozano
reponame_str Expeditio Repositorio Institucional UJTL
collection Expeditio Repositorio Institucional UJTL
bitstream.url.fl_str_mv https://expeditiorepositorio.utadeo.edu.co/bitstream/20.500.12010/20764/4/Anomalias_CRM.pdf.jpg
https://expeditiorepositorio.utadeo.edu.co/bitstream/20.500.12010/20764/1/Anomalias_CRM.pdf
https://expeditiorepositorio.utadeo.edu.co/bitstream/20.500.12010/20764/2/license.txt
https://expeditiorepositorio.utadeo.edu.co/bitstream/20.500.12010/20764/3/SingOGFOR-EFE-GDB-007_AUTORIZACION_DE_PUBLICACION_DE_TESIS_O_TRABAJO_DE_GRADO-signed.pdf
bitstream.checksum.fl_str_mv 3d77b1cd567326e2d8094e8709074d5d
601dd2e6d0dffdfddb7562807aabd4a8
baba314677a6b940f072575a13bb6906
50a873854afeb86f214a6cf60e3c3cdf
bitstream.checksumAlgorithm.fl_str_mv MD5
MD5
MD5
MD5
repository.name.fl_str_mv Repositorio Institucional - Universidad Jorge Tadeo Lozano
repository.mail.fl_str_mv expeditio@utadeo.edu.co
_version_ 1814213686630285312
spelling Colombia2021-07-29T15:33:03Z2021-07-29T15:33:03Z2021http://hdl.handle.net/20.500.12010/20764http://expeditio.utadeo.edu.coDeep learning models have improved several business intelligence tools like Customer relationship Management(CRM) systems. However, those models have increased the need for advanced computational capacity and infrastructure. Modern accelerators are starting to have floating-point precision arithmetic problems generated by highly streamlined systems, powered by the need to process an ever-increasing volume of data and increasingly complex models to attend to the necessity to identify customer data that allow consolidating products or services. We focus on CRM anomalies detection using GPU(Graphics Processor Unit) because they are a relevant source of money drain for organizations and directly affect the relationship between clients and suppliers. Our results present the combination of deep learning models with a computational structure that could access by organizations, but with a combination that reduces the number of features that achieve answers to CRM system.#AnáliticaDeDatosLos modelos de aprendizaje profundo han mejorado varias herramientas de inteligencia empresarial, como los sistemas de gestión de relaciones con el cliente (CRM). Sin embargo, esos modelos han aumentado la necesidad de infraestructura y capacidad computacional avanzada. Los aceleradores modernos están comenzando a tener problemas aritméticos de precisión de punto flotante generados por sistemas altamente optimizados, impulsados ​​por la necesidad de procesar un volumen cada vez mayor de datos y modelos cada vez más complejos para atender la necesidad de identificar datos de clientes que permitan consolidar productos o servicios. . Nos enfocamos en la detección de anomalías de CRM utilizando GPU (Graphics Processor Unit) porque son una fuente relevante de drenaje de dinero para las organizaciones y afectan directamente la relación entre clientes y proveedores. Nuestros resultados presentan la combinación de modelos de aprendizaje profundo con una estructura computacional a la que podrían acceder las organizaciones, pero con una combinación que reduce la cantidad de funcionalidades que logran respuestas al sistema CRM.12 páginasapplication/pdfengUniversidad de Bogotá Jorge Tadeo LozanoMaestría en Ingeniería y Analítica de Datosinstname:Universidad de Bogotá Jorge Tadeo Lozanoreponame:Expeditio Repositorio Institucional UJTLAnalítica de datosAnálisis de sistemas -- Tesis y disertaciones académicasTeoría de sistemas -- Tesis y disertaciones académicasDiseño de sistemas -- Tesis y disertaciones académicasUnsupervised learning models-based CRM anomaly detection using GPUTrabajo de grado de maestríainfo:eu-repo/semantics/masterThesishttp://purl.org/coar/resource_type/c_2df8fbb1Abierto (Texto Completo)http://purl.org/coar/access_right/c_abf2Akhtar, M.S., Chauhan, D.S., Ekbal, A., 2020. A deep multi-task contextual attention framework for multi-modal affect analysis. ACM Trans. Knowl. Discov. Data 14. doi:10.1145/3380744.Navarro, C.A., Carrasco, R., Barrientos, R.J., Riquelme, J.A., Vega, R., 2021. Gpu tensor cores for fast arithmetic reductions. IEEE Transactions on Parallel and Distributed Systems 32, 72–84. doi:10.1109/ TPDS.2020.3011893.Nevin, J., 1995. Relationship marketing and distribution channels: Exploring fundamental issues. Journal of the Academy of Marketing Science 23, 327–334. doi:10.1177/009207039502300413.Ordóñez, F., Roggen, D., 2016. Deep convolutional and lstm recurrent neural networks for multimodal wearable activity recognition. Sensors (Switzerland) 16. doi:10.3390/s16010115.Ozar, B., . Sql-server-first-responder-kit. https://github.com/ BrentOzarULTD/SQL-Server-First-Responder-Kit/blob/dev/sp_ ineachdb.sql.Pang, G., Shen, C., Cao, L., Hengel, A.V.D., 2021. Deep learning for anomaly detection: A review. ACM Comput. Surv. 54. doi:10.1145/ 3439950.Payne, A., Frow, P., 2005. A strategic framework for customer relationship management. Journal of Marketing 69, 167–176. doi:10. 1509/jmkg.2005.69.4.167.Racherla, P., Hu, C., 2008. ecrm system adoption by hospitality organizations: A technology-organization-environment (toe) framework. Journal of Hospitality&Leisure Marketing 17, 30–58. URL: https:// doi.org/10.1080/10507050801978372, doi:10.1080/10507050801978372.Raina, R., Madhavan, A., Ng, A., 2009. Large-scale deep unsupervised learning using graphics processors, in: Proceedings of the 26th International Conference On Machine Learning, ICML 2009, p. 110. doi:10.1145/1553374.1553486.Sadhu, V., Misu, T., Pompili, D., 2019. Deep multi-task learning for anomalous driving detection using can bus scalar sensor data, in: 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 2038–2043. doi:10.1109/IROS40897.2019. 8967753.Sagheer, A., Kotb, M., 2019. Unsupervised pre-training of a deep lstm-based stacked autoencoder for multivariate time series forecasting problems. Scientific Reports 9, 19038. URL: https://doi.org/ 10.1038/s41598-019-55320-6, doi:10.1038/s41598-019-55320-6.Trainor, K., Andzulis, J., Rapp, A., Agnihotri, R., 2014. Social media technology usage and customer relationship performance: A capabilities-based examination of social crm. Journal of Business Research 67, 1201–1208. doi:10.1016/j.jbusres.2013.05.002.Wang, R., Nie, K., Wang, T., Yang, Y., Long, B., 2020. Deep learning for anomaly detection, in: Proceedings of the 13th International Conference on Web Search and Data Mining, p. 894–896. doi:10.1145/3336191.3371876.Wirth, R., Hipp, J., 2000. Crisp-dm: Towards a standard process model for data mining, in: Proceedings of the 4th international conference on the practical applications of knowledge discovery and data mining, Springer-Verlag London, UK. pp. 29–39.Zachariadis, O., Satpute, N., Gómez-Luna, J., Olivares, J., 2020. Accelerating sparse matrix–matrix multiplication with gpu tensor cores. Computers & Electrical Engineering 88, 106848. doi:https://doi. org/10.1016/j.compeleceng.2020.106848.Zhang, S., Wu, P., 2019. High accuracy low precision qr factorization and least square solver on gpu with tensorcore. ArXiv abs/1912.05508.Zheng, Z., Pu, J., Liu, L., Wang, D., Mei, X., Zhang, S., Dai, Q., 2020. Contextual anomaly detection in solder paste inspection with multi-task learning. ACM Trans. Intell. Syst. Technol. 11. URL: https://doi.org/10.1145/3383261, doi:10.1145/3383261.Brereton, R.G., Lloyd, G.R., 2016. Re-evaluating the role of the mahalanobis distance measure. Journal of Chemometrics 30, 134–143. doi:https://doi.org/10.1002/cem.2779.Chen, I., Popovich, K., 2003. Understanding customer relationship management (crm): People, process and technology. Business Process Management Journal 9, 672–688. doi:10.1108/ 14637150310496758.Chen, Z.Y., Fan, Z.P., Sun, M., 2012. A hierarchical multiple kernel support vector machine for customer churn prediction using longitudinal behavioral data. European Journal of Operational Research 223, 461–472. doi:https://doi.org/10.1016/j.ejor.2012.06.040.Bastidas Betancourt, DanielGarcía-Bedoya, OlmerGranados, Oscar M.Magíster en Ingeniería y Analítica de DatosTHUMBNAILAnomalias_CRM.pdf.jpgAnomalias_CRM.pdf.jpgIM Thumbnailimage/jpeg4165https://expeditiorepositorio.utadeo.edu.co/bitstream/20.500.12010/20764/4/Anomalias_CRM.pdf.jpg3d77b1cd567326e2d8094e8709074d5dMD54open accessORIGINALAnomalias_CRM.pdfAnomalias_CRM.pdfVer documentoapplication/pdf1371071https://expeditiorepositorio.utadeo.edu.co/bitstream/20.500.12010/20764/1/Anomalias_CRM.pdf601dd2e6d0dffdfddb7562807aabd4a8MD51open accessLICENSElicense.txtlicense.txttext/plain; charset=utf-82938https://expeditiorepositorio.utadeo.edu.co/bitstream/20.500.12010/20764/2/license.txtbaba314677a6b940f072575a13bb6906MD52open accessSingOGFOR-EFE-GDB-007_AUTORIZACION_DE_PUBLICACION_DE_TESIS_O_TRABAJO_DE_GRADO-signed.pdfSingOGFOR-EFE-GDB-007_AUTORIZACION_DE_PUBLICACION_DE_TESIS_O_TRABAJO_DE_GRADO-signed.pdfLicenciaapplication/pdf368996https://expeditiorepositorio.utadeo.edu.co/bitstream/20.500.12010/20764/3/SingOGFOR-EFE-GDB-007_AUTORIZACION_DE_PUBLICACION_DE_TESIS_O_TRABAJO_DE_GRADO-signed.pdf50a873854afeb86f214a6cf60e3c3cdfMD53open access20.500.12010/20764oai:expeditiorepositorio.utadeo.edu.co:20.500.12010/207642021-07-29 23:01:33.512open accessRepositorio Institucional - Universidad Jorge Tadeo Lozanoexpeditio@utadeo.edu.coQXV0b3Jpem8gYWwgU2lzdGVtYSBkZSBCaWJsaW90ZWNhcyBVbml2ZXJzaWRhZCBkZSBCb2dvdMOhIEpvcmdlIFRhZGVvIExvemFubyBwYXJhCnF1ZSBjb24gZmluZXMgYWNhZMOpbWljb3MsIHByZXNlcnZlLCBjb25zZXJ2ZSwgb3JnYW5pY2UsIGVkaXRlIHkgbW9kaWZpcXVlCnRlY25vbMOzZ2ljYW1lbnRlIGVsIGRvY3VtZW50byBhbnRlcmlvcm1lbnRlIGNhcmdhZG8gYWwgUmVwb3NpdG9yaW8gSW5zdGl0dWNpb25hbApFeHBlZGl0aW8KCkV4Y2VwdHVhbmRvIHF1ZSBlbCBkb2N1bWVudG8gc2VhIGNvbmZpZGVuY2lhbCwgYXV0b3Jpem8gYSB1c3VhcmlvcyBpbnRlcm5vcyB5CmV4dGVybm9zIGRlIGxhIEluc3RpdHVjacOzbiBhIGNvbnN1bHRhciB5IHJlcHJvZHVjaXIgZWwgY29udGVuaWRvIGRlbCBkb2N1bWVudG8KcGFyYSBmaW5lcyBhY2Fkw6ltaWNvcyBudW5jYSBwYXJhIHVzb3MgY29tZXJjaWFsZXMsIGN1YW5kbyBtZWRpYW50ZSBsYQpjb3JyZXNwb25kaWVudGUgY2l0YSBiaWJsaW9ncsOhZmljYSBzZSBsZSBkZSBjcsOpZGl0byBhIGxhIG9icmEgeSBzdShzKSBhdXRvcihzKS4KCkV4Y2VwdHVhbmRvIHF1ZSBlbCBkb2N1bWVudG8gc2VhIGNvbmZpZGVuY2lhbCwgYXV0b3Jpem8gYXBsaWNhciBsYSBsaWNlbmNpYSBkZWwKZXN0w6FuZGFyIGludGVybmFjaW9uYWwgQ3JlYXRpdmUgQ29tbW9ucyAoQXR0cmlidXRpb24tTm9uQ29tbWVyY2lhbC1Ob0Rlcml2YXRpdmVzCjQuMCBJbnRlcm5hdGlvbmFsKSBxdWUgaW5kaWNhIHF1ZSBjdWFscXVpZXIgcGVyc29uYSBwdWVkZSB1c2FyIGxhIG9icmEgZGFuZG8KY3LDqWRpdG8gYWwgYXV0b3IsIHNpbiBwb2RlciBjb21lcmNpYXIgY29uIGxhIG9icmEgeSBzaW4gZ2VuZXJhciBvYnJhcyBkZXJpdmFkYXMuCgpFbCAobG9zKSBhdXRvcihlcykgY2VydGlmaWNhKG4pIHF1ZSBlbCBkb2N1bWVudG8gbm8gaW5mcmluZ2UgbmkgYXRlbnRhIGNvbnRyYQpkZXJlY2hvcyBpbmR1c3RyaWFsZXMsIHBhdHJpbW9uaWFsZXMsIGludGVsZWN0dWFsZXMsIG1vcmFsZXMgbyBjdWFscXVpZXIgb3RybyBkZQp0ZXJjZXJvcywgYXPDrSBtaXNtbyBkZWNsYXJhbiBxdWUgbGEgVW5pdmVyc2lkYWQgSm9yZ2UgVGFkZW8gTG96YW5vIHNlIGVuY3VlbnRyYQpsaWJyZSBkZSB0b2RhIHJlc3BvbnNhYmlsaWRhZCBjaXZpbCwgYWRtaW5pc3RyYXRpdmEgeS9vIHBlbmFsIHF1ZSBwdWVkYSBkZXJpdmFyc2UKZGUgbGEgcHVibGljYWNpw7NuIGRlbCB0cmFiYWpvIGRlIGdyYWRvIHkvbyB0ZXNpcyBlbiBjYWxpZGFkIGRlIGFjY2VzbyBhYmllcnRvIHBvcgpjdWFscXVpZXIgbWVkaW8uCgpFbiBjdW1wbGltaWVudG8gY29uIGxvIGRpc3B1ZXN0byBlbiBsYSBMZXkgMTU4MSBkZSAyMDEyIHkgZXNwZWNpYWxtZW50ZSBlbiB2aXJ0dWQKZGUgbG8gZGlzcHVlc3RvIGVuIGVsIEFydMOtY3VsbyAxMCBkZWwgRGVjcmV0byAxMzc3IGRlIDIwMTMsIGF1dG9yaXpvIGEgbGEKVW5pdmVyc2lkYWQgSm9yZ2UgVGFkZW8gTG96YW5vIGEgcHJvY2VkZXIgY29uIGVsIHRyYXRhbWllbnRvIGRlIGxvcyBkYXRvcwpwZXJzb25hbGVzIHBhcmEgZmluZXMgYWNhZMOpbWljb3MsIGhpc3TDs3JpY29zLCBlc3RhZMOtc3RpY29zIHkgYWRtaW5pc3RyYXRpdm9zIGRlCmxhIEluc3RpdHVjacOzbi4gRGUgY29uZm9ybWlkYWQgY29uIGxvIGVzdGFibGVjaWRvIGVuIGVsIGFydMOtY3VsbyAzMCBkZSBsYSBMZXkgMjMKZGUgMTk4MiB5IGVsIGFydMOtY3VsbyAxMSBkZSBsYSBEZWNpc2nDs24gQW5kaW5hIDM1MSBkZSAxOTkzLCBhY2xhcmFtb3MgcXVlIOKAnExvcwpkZXJlY2hvcyBtb3JhbGVzIHNvYnJlIGVsIHRyYWJham8gc29uIHByb3BpZWRhZCBkZSBsb3MgYXV0b3Jlc+KAnSwgbG9zIGN1YWxlcyBzb24KaXJyZW51bmNpYWJsZXMsIGltcHJlc2NyaXB0aWJsZXMsIGluZW1iYXJnYWJsZXMgZSBpbmFsaWVuYWJsZXMuCgpDb24gZWwgcmVnaXN0cm8gZW4gbGEgcMOhZ2luYSwgYXV0b3Jpem8gZGUgbWFuZXJhIGV4cHJlc2EgYSBsYSBGVU5EQUNJw5NOIFVOSVZFUlNJREFECkRFIEJPR09Uw4EgSk9SR0UgVEFERU8gTE9aQU5PLCBlbCB0cmF0YW1pZW50byBkZSBtaXMgZGF0b3MgcGVyc29uYWxlcyBwYXJhIHByb2Nlc2FyCm8gY29uc2VydmFyLCBjb24gZmluZXMgZXN0YWTDrXN0aWNvcywgZGUgY29udHJvbCBvIHN1cGVydmlzacOzbiwgYXPDrSBjb21vIHBhcmEgZWwKZW52w61vIGRlIGluZm9ybWFjacOzbiB2w61hIGNvcnJlbyBlbGVjdHLDs25pY28sIGRlbnRybyBkZWwgbWFyY28gZXN0YWJsZWNpZG8gcG9yIGxhCkxleSAxNTgxIGRlIDIwMTIgeSBzdXMgZGVjcmV0b3MgY29tcGxlbWVudGFyaW9zIHNvYnJlIFRyYXRhbWllbnRvIGRlIERhdG9zClBlcnNvbmFsZXMuIEVuIGN1YWxxdWllciBjYXNvLCBlbnRpZW5kbyBxdWUgcG9kcsOpIGhhY2VyIHVzbyBkZWwgZGVyZWNobyBhIGNvbm9jZXIsCmFjdHVhbGl6YXIsIHJlY3RpZmljYXIgbyBzdXByaW1pciBsb3MgZGF0b3MgcGVyc29uYWxlcyBtZWRpYW50ZSBlbCBlbnbDrW8gZGUgdW5hCmNvbXVuaWNhY2nDs24gZXNjcml0YSBhbCBjb3JyZW8gZWxlY3Ryw7NuaWNvIHByb3RlY2Npb25kYXRvc0B1dGFkZW8uZWR1LmNvLgoKTGEgRlVOREFDScOTTiBVTklWRVJTSURBRCBERSBCT0dPVMOBIEpPUkdFIFRBREVPIExPWkFOTyBubyB1dGlsaXphcsOhIGxvcyBkYXRvcwpwZXJzb25hbGVzIHBhcmEgZmluZXMgZGlmZXJlbnRlcyBhIGxvcyBhbnVuY2lhZG9zIHkgZGFyw6EgdW4gdXNvIGFkZWN1YWRvIHkKcmVzcG9uc2FibGUgYSBzdXMgZGF0b3MgcGVyc29uYWxlcyBkZSBhY3VlcmRvIGNvbiBsYSBkaXJlY3RyaXogZGUgUHJvdGVjY2nDs24gZGUKRGF0b3MgUGVyc29uYWxlcyBxdWUgcG9kcsOhIGNvbnN1bHRhciBlbjoKaHR0cDovL3d3dy51dGFkZW8uZWR1LmNvL2VzL2xpbmsvZGVzY3VicmUtbGEtdW5pdmVyc2lkYWQvMi9kb2N1bWVudG9zCg==