Introduction
This is a collection of Matlab 6 functions developed for the PhD of Victor
J.A.S. Lobo.
General purpose classification routines
- knn.m - Classifies a data pattern according to the KNN
rule (k-nearest-neighbour)
- knn_mat.m - Classifies a number of samples according
to the KNN rule
- SelfClassify.m - Classify all patterns of a
dataset using the other patterns and knn
Q-Set related routines
- qs_mat_build - Builds Q-sets for a set of
patterns
- qs_memory_build - Builds Q-sets for a set of
patterns (memory efficient version)
- qsgc_mat_build - Build the generalized Q-Sets
for a set of patterns
- qs_build.m - Builds the Q-set for one given pattern
- g2p.m - Genertes Q-sets (positive only), from
generalized Q-Sets
- qs_select_heuristic.m - Selects a small
number of classifying prototypes using a heuristic method
- qs_select.m - Selects the minimum number of
classifying prototypes using a branch-and-bound
Protototype selection routines
- cnn.m - Condensed Nearest Neighnors
- rnn.m - Reduced Nearest Neighbors
Graphics routines (for plotting 2D data and prototypes)
- voronoi_boundary.m - Plots the voronoi
boudaries between two sets of classifier prototypes
- class_plot.m - Plots different classes using
different colors
Other routines
- confusionMatrix.m - Computes the
confusion matrix of a given dataset
- discriminant.m - Computes the discriminant for
data with a gaussian distribution
- splitData.m - Split a dateset for "leave-n-out"
cross-validation
- buildTrainTestSet.m - Build train-and-test
sets from split data
- read_koh.m - Read a SOMPAK datafile into matlab
- write_koh.m - Write a SOMPAK datafile
- Remove_col.m - Remove a column from a matlab
matrix
- spectra_wavfile.m - Generate spectra from
WAV files using the desired parameters
- mHamming.m - Modified (corrected) Hamming sampling
window
- hann.m - Von Hann sampling window
- hauddorff.m - Computes the Hausdorff distance
between vectors
- plot_haus.m - Computes the Hausdorff distance
between vectors, plots the difference between the vectors, and the plots the
computed difference
- FindPrimes.m - Find prime implicants of a boolean
function
- Batch files
- PERIOD.M - Batch file that computes a spectogram
of a WAV file, and extracts binary patterns from it
- mk_compare_nn_cnn_rnn_lnn.m -
Batch file that compares Nearest Neighbors, Condensed Nearest Neighbors,
Reduced Nearest Neighbors, and Q-Sets (Lobo Nearest Neighbors) for the
double F problem.
Routines for particular datasets