Mutual Information

Calculates the mutual information between two discrete variables (or a group and a single variable).

Description

MutualInformation: returns mutual information (in bits) of discrete variables 'X' and 'Y'

  I = MutualInformation(X,Y);
 
  I = calculated mutual information (in bits)
  X = variable(s) to be analyzed (column vector)
  Y = variable to be analyzed (column vector)
 
  Note 1: Multiple variables may be handled jointly as columns in
    matrix 'X'.
  Note 2: Requires the 'Entropy' and 'JointEntropy' functions.

• MATLAB release: MATLAB 6.0 (R12)
• Other requirements: Requires JointEntropy and Entropy functions.

[Marcello Costa]

Good Job

[Jun]

I am not sure if I run the code right or there is a bug in the code.

I ran the following and got the answers below. Does not seem right to me. A and B are completely independent random numbers, which I verified.
 
>> A=rand(1,100);
>> B=rand(1,100);
>> MutualInformation(A',A')

ans =

    6.6439

>> MutualInformation(B',B')

ans =

    6.6439

>> MutualInformation(A',B')

ans =

    6.6439

[Will Dwinnell]

This routine is intended for use with discrete variables, not continuous ones.