1 / 17

Satellital Image Clasification with neural networks Step implemented – Final Report

www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008. Satellital Image Clasification with neural networks Step implemented – Final Report Susana Arias, Héctor Gómez UNIVERSIDAD TÉCNICA PARTICULAR DE LOJA ECUADOR.

tyler
Télécharger la présentation

Satellital Image Clasification with neural networks Step implemented – Final Report

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Satellital Image Clasification with neural networks Step implemented – Final Report Susana Arias, Héctor Gómez UNIVERSIDAD TÉCNICA PARTICULAR DE LOJA ECUADOR

  2. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report • Satellital Image Clasification use a neural network for clasification the percentage of land use of a satellite image.

  3. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report First Solution • We decide to compiler the windows's application in Linux, for that We use a mono's program (C# interface for Linux). 1. Create info about mono repository: cd /etc/yum.repos.d 2. Download and install: wget http://www.go-mono.com/download-stable/rhel-4-i386/mono.repo 3. We sure Linux kernel can run Mono applications "natively" (like Windows), without using "mono" every time. Following these directions, We create a script.

  4. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report • Problems encountered: Execute the commands only like root. [alumno@m150-e2gris1 rn]$ ./monos ./monos: line 18: /proc/sys/fs/binfmt_misc/register: Permission denied [alumno@m150-e2gris1 rn]$ sudo ./monos • Solution Install mono´s program in the worker node

  5. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report Second Solution • We translate the application from .net to C language with MPI libraries.

  6. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report Second Solution file bf.c #include <stdlib.h> #include <stdio.h> #include <math.h> #include "mpi.h" /********** M A I N ******************************************************************************/ int numprocesses, procnum; int main(int nargs, char* argv[]) { NET Net; BOOL Stop; int result=MPI_Init(&nargs, &argv); MPI_Comm_size(MPI_COMM_WORLD, &numprocesses); // get total number of processes MPI_Comm_rank(MPI_COMM_WORLD, &procnum); // get process number for this process

  7. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report • Then we compiler MPI en UI, $ mpiCC • Compiler our application, $ mpiCC -o bf bf.c • Then create a .jdl Type = “Job”; JobType = “MPICH”; Executable = “bf”; CPUNumber = 10; StdOutput = “output”; StdError = “error”; InputSandbox = {“bf”}; OutputSandbox = {“BPN.txt”};

  8. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report • We compiler bf.c ONLY in the 64 bits machine, do not compiler the machine in 32 bits because do not have installed MPI. [app01@m111-e2gris1 BF]$ glite-wms-job-submit -a -o jid bf.jdl [app01@m111-e2gris1 BF]$ glite-wms-job-status -i jid [app01@m111-e2gris1 BF]$ cd /tmp/jobOutput/app01_776TAA4JvsR-NP2gRkgamA/ BPN.txt NMSE is 0.106 on Training Set and 0.133 on Test Set NMSE is 0.105 on Training Set and 0.129 on Test Set NMSE is 0.104 on Training Set and 0.128 on Test Set

  9. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report • Then the application use a Storage Element for read the input files (Satellital Image) and run in the worker nodes. JOB GET OUTPUT OUTCOME Output sandbox files for the job: https://m05-e2gris1.eela.ufrj.br:9000/aUZWfM3V6-ftNkvfdG3MGA have been successfully retrieved and stored in the directory: /tmp/jobOutput/app01_aUZWfM3V6-ftNkvfdG3MGA [app01@m111-e2gris1 app01_aUZWfM3V6-ftNkvfdG3MGA]$ ls BPN.txt error output watchdog.log

  10. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report • Finally We use a GENIUS portal for submit the jobs, not with MPI.

  11. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report: MrBayes As the same time, We worked in another application like Mrbayes (bionformatics), We install mrbayes at the UI, and We try to run mrbayes with condor, but it was not possible because there was not condor in the CE, finally We saw how to mrbyes runs with mpi and find out how to put some directives to run in the WN.

  12. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Final Report: MrBayes • We send MrBayes application to GENIUS portal. • Normally the time for run is 2 weeks, in 1 CPU, and the output files is a string sequence of 360GB

  13. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008

  14. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Thanks!

  15. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Questions?

  16. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Questions?

  17. www.eu-eela.eu Itacuruça – Brazil, november 2 -15, 2008 Thanks!

More Related