Classer Script


ClasserScript Interface to Classer Toolkit:

image

ClasserScript is a batch interface to the Classer toolkit. More specifically, ClasserScript is a miniature language; it lets you define views of data, classifier models, and iterative combinations of the two to generate a variety of output metrics.

ClasserScript offers a simple environment for doing ARTMAP-based research without having to write any code. It lets you script simulations, preprocess data, and generate performance metrics that can be used in tuning an ARTMAP classifier.

ClasserScript Software:
The ClasserScript executable, csd, ("ClasserScript Driver") is available here.

Documentation
The
ClasserScript User Guide (v1.1) is a comprehensive introduction to using ClasserScript, and more generally, to the Classer toolkit.

The User's Guide includes:

  • A detailed description of the command syntax for ClasserScript
  • A discussion of data set formats used by Classer
  • Examples of Classer's graphical outputs
  • An introduction to the data sets that are made available with Classer
  • A reference section summarizing ClasserScript's syntax.
Scripts Cookbook:
The following set of scripts provides a cookbook of examples to help you get started with ClasserScript's basic functionality. The data sets must be downloaded before the scripts can be used. In addition, unless you install the data sets at the top directory level, you will need to rename their locations in the scripts.

Script Name Data set Notes
loadData Boston Demonstrates advantage of binary representation for large data sets
showData Boston Generates six images from the data set, storing them in bosImg?.ppm
basicPrints CIS Prints memory size, timing, number of F2 nodes
printCmat Letters (CV) Good performance (96.9%), 5x5 cross-validation, confusion matrix printout
foreach CIS (CV) Prints timing and percent correct for 100 vigilance values
foreach2D Letters (CV) Tabular printout, exploring 2D parameter space
setFeatures(1) Letters Features 1-11, poor performance (78%)
setFeatures(2) Letters Features 3-13, medium performance (49%)
setFeatures(3) Letters Features 6-16, good performance (92%)
setVigilance CIS (CV) 3 vigilance values, increasing performance & memory
setVoters CIS (CV) 1, 3, 5 voters, increasing performance & memory
outputCmat Letters Graphical confusion matrix ᎂ cmPlot.pgm
outputPred Letters Graphical prediction plot ᎂ lettersPred.ppm
thematic Boston Train GT, classify all, fuzzy ARTMAP, cap=268 ᎂ mapThematic.ppm
bosCv Boston Demonstrates cross-validation with user-specified partitions