Deriving labels from probabilities

If probabilities are provided, how do we derive labels from them in an LDA? What is the relationship between labels and probabilities?

[Walter Roberson]

Please show your code. The routines that are named LDA are in the File Exchange; the Linear Discriminant Analysis routine available from Mathworks is named something completely different.

[Lester Lim]

Im using this to get my labels: http://www.mathworks.com/matlabcentral/fileexchange/29673-lda-linear-discriminant-analysis

Another probelm I faced was I got probabilities less than 0.01... Any suggestions?