This was your last question. Congratulations!

Is there anything you want us to consider when grading your test, or do you have any feedback? 

(If not, it will help us if you just leave the field empty, and not write "No" or similar ).

The UCB1 algorithm uses the following formula when determining which arm to pull in the Multi-Armed Bandit problem:

Which of the following describes best the role of the parameter y:

The "Zero" in AlphaZero derives from the fact that the algorithm learns without playing a single game against itself.

Which of the following components is not part of the definition for Machine Learning by Tom Mitchell?

The key idea of backpropagation for multi-layer neural networks is to use a weighted combination of the error signals of all its predecessor nodes for the update in an interior node of the network.

Below you see a schematic input image (left), a convolution operator (middle), the resulting output image.

What value will be computed for the pixel marked with a question mark in the output?

Zermelo's Retrograde Analysis algorithm, which we discussed in the lecture, can solve any deterministic zero-sum perfect information game provided there is enough memory and computation time. 

According to the taxonomy introduced in the lecture, the resulting solution is

The graph below shows a simple game of chance where the MAX player can select between two fair coins, with different win expecations.

Which coin should MAX select in order to maximize her expected long-term reward?

Which of the following best describes Monte-Carlo Search?