| DEMO_LINCLASS | |
Demo on the algorithms learning linear classifiers.
Synopsis:
demo_linclass
Description:
DEMO_LINCLASS demonstrates use of the algorithms which find
linear decision hyperplane between two (dichotomy)
vector sets. The demo requires 2D input training data.
The program vizualizes found hyperplane in the current
algorithm step. The missclassified vector used by the
demonstrated iterative algorithms for update is vizualized
as well. Text description of the found solution is
printed at the bottom part of window.
Following algorithms can be tested:
Perceptron - Perceptron learning rule (see 'help perceptron').
Kozinec - Kozinec's algorithm (see 'help ekozinec', eps=-1).
e-Kozinec - Kozinec's algorithm finding eps-optimal hyperplane
(see 'help ekozinec', eps > 0).
Linear SVM - Linear Supprot Vector Machines for separable data
(see 'help smo', C=inf, ker='linear').
Control:
Algorithm - Dselects algorithm for testing.
Epsilon - Input parameter of 'ekozinec' algorithm
(see 'help ekozinec').
Iterations - Number of iterations in one step.
Animation - Enables/dissables animation - smooth changeover
between two algorithm states.
FIG2EPS - Exports screen to the PostScript file.
Load data - Loads input training data from file.
Create data - Invokes program for creating training data.
Reset - Sets the algorithm to the initial state.
Play - Runs the algorithm.
Stop - Stops the running algorithm.
Step - Perform one step of the algorithm.
Info - Invoke the info box.
Close - Close the program.
See also
PERCEPTRON, EKOZINEC, SVM.
About: Statistical Pattern Recognition Toolbox
(C) 1999-2003, Written by Vojtech Franc and Vaclav Hlavac
Czech Technical University Prague
Faculty of Electrical Engineering
Center for Machine Perception
Modifications:
19-sep-2003, VF
17-Feb-2003, VF
24. 6.00 V. Hlavac, comments polished.
11-dec-2000 V. Franc, a little increasing of code readibility
15-dec-2000