Answer & Explanation:

Question 1

• Correct Answer: D) It randomly deactivates a percentage of neurons per iteration, forcing the network to learn redundant representations.

• Explanation: Dropout randomly drops units during training, which prevents neurons from co-adapting too much. Because a neuron cannot rely on the presence of specific other neurons, it is forced to learn more robust features that are useful in conjunction with many different random subsets of other neurons.

Question 2

• Correct Answer: C) L1 uses an absolute value penalty whose constant derivative creates sharp corners on the axis boundaries.

• Explanation: The budget constraint of L1 regularization forms a diamond shape (in 2D space) with sharp corners on the coordinate axes. When the contours of the unregularized loss function meet this diamond constraint, they are highly likely to hit one of these sharp corners first, setting the weight of that axis to exactly zero and producing a sparse model.