Probabilistic & Margin-Based Methods — Checkpoint Quiz

B is correct. Naive Bayes multiplies together each feature's individual likelihood as if features don't interact at all, given the class. That independence assumption is almost never exactly true, but the model is often useful anyway.

B is correct. SVM looks for the decision boundary that maximizes this gap, the margin, on the theory that a wider margin generalizes better to new data.

B is correct. Kernels compute what the dot product between two points would be in a higher-dimensional space, without ever actually constructing that space, which makes non-linear boundaries computationally feasible.