Once the applet is loaded, and this could take a while, you will notice three toolbars and a data canvas appearing in the main applet frame.

Main Applet Frame

Action toolbar is located at the top of the frame and is used to apply specific classifiers to existing data sets.

	Exit the applet.
	Run the Brute Force algorithm.
	Run the Neural Network.
	Run the Genetic Algorithm.
	About this Applet.

Data toolbar is located at the left side of the frame and is used to handle data sets.

	Clear data sets.
	Create/Update data sets.

Parameter toolbar is located at the right side of the frame and is used to change parameters affecting performance of different classifiers.

	Restore default values of all parameters.
	Brute Force algorithm parameters.
	Neural Network parameters.
	Genetic Algorithm parameters.

Data canvas will show data points and discriminant functions which are color coded.

Green	Brute Force.
Blue	Neural Network.
Red	Genetic Algorithm.

Data Distribution and Creation Frame

From this frame you control creation of data points which are to be classified.

A specified number of events from either class (x or y) is generated according to a particular distribution (Gaussian or flat).

Distributions are parameterized by the mean and the three elements of the covariance matrix.

A bivariate Gaussian is defined in the usual way while the flat distribution has uniform probability density over an ellipse (points which are at a unit Mahalanobis distance from the mean) defined by the covariance matrix.

Rotation buttons can be used to observe the 3D graph of the probability density from different view points.

Parameter Selection Frames

These dialogs are used to set the parameters used by different classifiers, as discussed in the previous section. Use the sliders to pick a value from the allowed range. Reasonable defaults are provided.

Final Hints

You must start by creating data sets, which can after be updated. Both classifiers which require training (GA and NN) can be retrained to improve the performance. Each consecutive execution of an action tool will train the current classifier through an additional number of epochs. To start with a new (random) classifier you must either clear the data set or restore default values from the parameter toolbar.

View applet documentation created by javadoc.
To speed up the execution of the applet download the classes, unzip the file and start the applet by java CApplet.