pkfsann - feature selection for nn classifier

       Classification  problems  dealing  with  high dimensional input data can be challenging due to the Hughes
       phenomenon.  Hyperspectral data, for instance, can have hundreds of spectral bands  and  require  special
       attention when being classified.  In particular when limited training data are available, the classifica‐
       tion of such data can be problematic without reducing the dimension.

       pkfsann  implements  a  number of feature selection techniques, among which a sequential floating forward
       search (SFFS).  Also consider the SVM classifier implemented in pksvm(1), which has been shown to be more
       robust to this type of problem than others.

       pkfsann - feature selection for nn classifier

-tfilename, --trainingfilename
              training vector file.  A single vector file contains all training features (must be  set  as:  B0,
              B1,  B2,...)  for  all  classes (class numbers identified by label option).  Use multiple training
              files for bootstrap aggregation (alternative to the bag and bsize options, where a  random  subset
              is taken from a single training file)

       -nnumber, --nfnumber
              number of features to select (0 to select optimal number, see also --ecost option)

       -ifilename, --inputfilename
              input test set (leave empty to perform a cross validation based on training only)

       -vlevel, --verboselevel
              set to: 0 (results only), 1 (confusion matrix), 2 (debug)

       Advanced options

       -tlnlayer, --tlnlayer
              training layer name(s)

       -labelattribute, --labelattribute
              identifier for class label in training vector file.  (default: label)

       --balancesize
              balance the input data to this number of samples for each class (default: 0)

       -random, --random
              in case of balance, randomize input data

       -minnumber, --minnumber
              if number of training pixels is less then min, do not take this class into account

       -bband, --bandband
              band index (starting from 0, either use band option or use start to end)

       -sbandband, --startbandband
              start band sequence number

       -ebandband, --endbandband
              end band sequence number

       -offsetvalue, --offsetvalue
              offset value for each spectral band input features: refl[band]=(DN[band]-offset[band])/scale[band]

       -scalevalue, --scalevalue
              scale value for each spectral band input features: refl=(DN[band]-offset[band])/scale[band] (use 0
              if scale min and max in each band to -1.0 and 1.0)

       -a0|1|2, --aggreg0|1|2
              how  to  combine  aggregated classifiers, see also --rc option (0: no aggregation, 1: sum rule, 2:
              max rule).

       -smmethod, --smmethod
              feature selection method (sffs=sequential floating forward search, sfs=sequential forward  search,
              sbs, sequential backward search, bfs=brute force search)

       -ecostvalue, --ecostvalue
              epsilon for stopping criterion in cost function to determine optimal number of features

       -cvvalue, --cvvalue
              n-fold cross validation mode (default: 0)

       -cname, --classname
              list of class names.

       -rvalue, --reclassvalue
              list of class values (use same order as in --class option).

       -nnumber, --nneuronnumber
              number  of  neurons in hidden layers in neural network (multiple hidden layers are set by defining
              multiple number of neurons: -nn 15 -nn 1, default is one hidden layer with 5 neurons)

       --connection0|1
              connection rate (default: 1.0 for a fully connected network)

       -wweights, --weightsweights
              weights for neural network.  Apply to fully connected network only, starting from first input neu‐
              ron to last output neuron, including the bias neurons (last neuron in each but last layer)

       -lrate, --learningrate
              learning rate (default: 0.7)

       --maxitnumber
              number of maximum iterations (epoch) (default: 500)

pksvm(1)

                                                 01 January 2025                                      pkfsann(1)

pkfsann-ttraining-nnumber [options] [advancedoptions]