#likelihood

Question 1: Single choice

Maximum likelihood estimation chooses $\theta$ to maximize $\prod_n p_\theta(y^{(n)} \mid x^{(n)})$ over the training set. In this likelihood, what is held fixed and what varies?

The observed examples are fixed, and it is a function of $\theta$ 1
The parameters $\theta$ are fixed, and it is a function of the targets $y$ 2
The data are fixed, and it is a probability distribution over $\theta$ 3
The data are fixed, and it gives the probability that $\theta$ is the true value4

Timed test