Fuzzy k-NN

Fuzzy k-nearest neighbors classifier that was proposed by J. M. Keller et.al.

Fuzzy k-nearest neighbors classifier that can work with training samples whose label info is fuzzified. The prototype is as follows.

[y,predict_class] = f_knn(tr,tr_memberships,te,k)

tr: training samples
tr_memberships: membership values of traning samples
te: testing samples
k: k value vector (more than one value is possible)
y: output memberships for testing set
predict_class: the most likely classes for testing set