Yaniv ran us through our second session on EM algorithms. We implemented a simple case described in this tutorial.
[gist id=1028113]
Code doesn’t reflect the abstraction of the algorithm into a proper Expectation step and Maximization step. We attempted this generalization:
[gist id=1028122]
Missed something in framing this correctly, since the maximization step includes a function that doesn’t depend on s[1]. Any ideas?