OrgLion-ML
/
dlib
mirror of https://github.com/davisking/dlib.git
1
0
Fork 0
Code Issues Releases Wiki Activity

Updated the rank features example to discuss the newer kcentroid 

max sv argument.

--HG--
extra : convert_revision : svn%3Afdd8eb12-d10e-0410-9acb-85c331704f74/trunk%402938
This commit is contained in:
Davis King 2009-03-16 02:58:47 +00:00
parent 40050a2870
commit 43a7f59054
1 changed files with 20 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
examples/rank_features_ex.cpp
View File

@ -97,23 +97,29 @@ int main()
// classification and feature ranking. // classification and feature ranking.
typedef radial_basis_kernel<sample_type> kernel_type; typedef radial_basis_kernel<sample_type> kernel_type;
// This line here declares the kcentroid object we want to use for feature ranking. Note that there // Here we declare an instance of the kcentroid object. It is used by rank_features()
// are two numbers in it. The first is the argument to the kernel. The second is a tolerance argument // two represent the centroids of the two classes. The kcentroid has 4 parameters
// for the kcentroid object. This tolerance is basically a control on the number of support vectors it // you need to set. The first argument to the constructor is the kernel we wish to
// will use, with a smaller tolerance giving better accuracy but longer running times. Generally // use. The second is a parameter that determines the numerical accuracy with which
// something in the range 0.01 to 0.001 is a good choice. // the object will perform part of the ranking algorithm. Generally, smaller values
kcentroid<kernel_type> kc(kernel_type(0.05), 0.001); // give better results but cause the algorithm to attempt to use more support vectors
// (and thus run slower and use more memory). The third argument, however, is the
// maximum number of support vectors a kcentroid is allowed to use. So you can use
// it to control the complexity. Finally, the last argument should always be set to
// false when using a kcentroid for ranking (see the kcentroid docs for details on
// this parameter).
kcentroid<kernel_type> kc(kernel_type(0.05), 0.001, 25, false);
// And finally we get to the feature ranking. Here we call rank_features() with the kcentroid we just made, // And finally we get to the feature ranking. Here we call rank_features() with the kcentroid we just made,
// the samples and labels we made above, and the number of features we want it to rank. // the samples and labels we made above, and the number of features we want it to rank.
cout << rank_features(kc, samples, labels, 4) << endl; cout << rank_features(kc, samples, labels) << endl;
// The output is: // The output is:
/* /*
1 0.514169 1 0.514254
0 0.810535 0 0.810668
3 1 3 1
2 0.966936 2 0.994169
*/ */
// The first column is a list of the features in order of decreasing goodness. So the rank_features() function // The first column is a list of the features in order of decreasing goodness. So the rank_features() function
@ -128,10 +134,10 @@ int main()
// indicate a larger separation. // indicate a larger separation.
// So to break it down a little more. // So to break it down a little more.
// 1 0.514169 <-- class separation of feature 1 all by itself // 1 0.514254 <-- class separation of feature 1 all by itself
// 0 0.810535 <-- class separation of feature 1 and 0 // 0 0.810668 <-- class separation of feature 1 and 0
// 3 1 <-- class separation of feature 1, 0, and 3 // 3 1 <-- class separation of feature 1, 0, and 3
// 2 0.966936 <-- class separation of feature 1, 0, 3, and 2 // 2 0.994169 <-- class separation of feature 1, 0, 3, and 2
} }