EM M-Step Mean Update

Problem

In the M-step of EM for a Gaussian mixture, the updated mean of the given component is the responsibility-weighted average of the data: the sum of responsibilities[n][component] * data[n] divided by the sum of those responsibilities. Return 0.0 when the weights sum to zero or the shapes are invalid.

Hint

This is a weighted average where the responsibilities are the weights.

Test cases

case 1

in: [[1,2,3,4],[[0.9,0.1],[0.8,0.2],[0.2,0.8],[0.1,0.9]],0]

expected: 1.75

case 2

in: [[1,2,3,4],[[0.9,0.1],[0.8,0.2],[0.2,0.8],[0.1,0.9]],1]

expected: 3.25

case 3

in: [[1,2],[[1],[1]],0]

expected: 1.5

case 4

in: [[1,2],[[0],[0]],0]

expected: 0

case 5

in: [[],[],0]

expected: 0

Starter codePython

def em_m_step_mean(data, responsibilities, component):
    # Your code here
    pass

5 test cases · run your code against them in the editor