Skip to content

Latest commit

 

History

History
20 lines (17 loc) · 1.3 KB

README.md

File metadata and controls

20 lines (17 loc) · 1.3 KB
Raw

FastEE

FastEE: Fast Ensembles of Elastic Distances This is the source code for FastEE - Faster version of the Ensembles of Elastic Distances (EE). In particular, FastEE tackles the long training time of EE. This code only focus on training EE.

Running the code:

Running from terminal

  1. Training the individual classifers
  • java -Xmx14g -Xms14g -cp $LIBDIR: experiments.IndividualClassifierEE $OUTPUTDIR $DATASETDIR $PROBLEM $DISTANCE
  • java -Xmx14g -Xms14g -cp $LIBDIR: experiments.IndividualClassifierLbEE $OUTPUTDIR $DATASETDIR $PROBLEM $DISTANCE
  • java -Xmx14g -Xms14g -cp $LIBDIR: experiments.IndividualClassifierFastEE $OUTPUTDIR $DATASETDIR $PROBLEM $DISTANCE
  • java -Xmx14g -Xms14g -cp $LIBDIR: experiments.IndividualClassifierApproxEE $OUTPUTDIR $DATASETDIR $PROBLEM $DISTANCE $NSAMPLES $NRUNS
  1. Training the whole ensemble
  • java -Xmx14g -Xms14g -cp $LIBDIR: experiments.TrainElasticEnsembles $OUTPUTDIR $DATASETDIR $PROBLEM $CLASSIFIER $NSAMPLES

Running from Bash Script

  1. bash TrainIndividualClasssifiers.sh [-p <Dataset_Name>] [-c <EE|LbEE|FastEE|ApproxEE>] [-d <Dataset_Directory>] [-o <Output_Directory>] [-s <Number_of_Samples>] [-r <Number_of_Runs>]
  2. bash TrainEnsembles.sh [-p <Dataset_Name>] [-c <EE|LbEE|FastEE|ApproxEE>] [-d <Dataset_Directory>] [-o <Output_Directory>] [-s <Number_of_Samples>]