For the description of the divergence you can take a look at :
I ask if I can use it instead of Euclidean distance
Thank you for submitting this, however I was wondering what is the reason for the line: P = P ./repmat(sum(P,2),[1 size(P,2)]);?
This doesn't seem to be a part of other KLdiv calculations.
It will be better to give an example
I have two probability matrix P and Q of size m x n and q x n each. How to calculate KL-divergence for these two matrices?
Shouldn't the log value be log2 instead of natural logarithm?
Sir, can you give a detailed description of P and Q in this function? example better.
dear sir we need kld for feature selection in MRI brain image can some one help me
Please note that the logarithms should be base 2 if calculating values in bits.
Your implementation is third of its kind on FEX: please take time to search FEX for existing code before posting.
Download apps, toolboxes, and other File Exchange content using Add-On Explorer in MATLAB.