Adding correction to momentum in week 2 assignment

[martiva]

I just thought I’d share that I checked what the result would be if I add the correction for the momentum optimization method in the week 2 assignment. The result was 88% accuracy (compared to 71% without correction). I thought this is interesting as it significantly increases the accuracy compared to normal gradient descent and also gets very close to the performance of Adam opt method :).

[TMosh]

Thanks for your report!