DEMO_ANDERSON |
Demo on Generalized Anderson's task.
Synopsis:
demo_anderson
Description:
This demo demonstrates the algorithms which solve
the Generalized Anderson`s Task (GAT) [SH10]. The GAT is an
instance of the non-Bayesian task of decision under
non-random intervention.
The goal of is to find a binary linear classification
rule (g(x)=sgn(W'*x+b) (line in 2D) with minimal probability of
misclassification. The conditional probabilities are known to
be Gaussians their paramaters belong to a given set of
parameters. The true parameters are not known. The linear rule
which guarantes the minimimal classification error for the worst
possible case (the worst configuration of Gaussains) is
sought for.
The found solution (hyperplane, line in 2D) is vizualized
as well as the input Gaussians which describe input classes.
Following algorithms can be tested:
Eps-solution - Finds epsilon-solution of the GAT in finite number
of iterations if such solution exist. The epsilon means
desired classification error.
Original - Original Anderson-Bahadur's algorithm defined for
two Gaussians only (each class one Gaussian).
Optimal - Implementation of general algorithm propsed by Schlesinger.
It finds the optimal solution.
Gradient - Fast and simple implementation which uses the generalized
gradient descent optimization.
Control:
Algorithm - select algorithm for testing.
Parameter - parameters for the selected algorithm.
Iterations - number of iterations in one step.
Animation - enable/dissable animation.
FIG2EPS - export screen to the PostScript file.
Load data - load input point sets from file.
Create data - call interactive program for creating sets of Gaussians.
Reset - set the tested algorithm to the initial state.
Play - run the tested algorithm.
Stop - stop the running algorithm.
Step - perform only one step.
Info - display the info box.
Close - close the program.
See also
EANDERS, ANDRORIG, GGRADANDR, GANDERS.
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:
17-sep-2003, VF
11-June-2001, V.Franc, comments added.
24. 6.00 V. Hlavac, comments polished.