1.
Using the k‐NN algorithm to model how people perceive similarity between them faces the problem that each instance is valued the same although similarity estimates can be asymmetric (Tversky, 1977). A smart approach could be to assign weights to the contributions of the neighbors, so that the nearer neighbor adds more to the average than distant ones (e.g., 1 divided by the distance to the neighbor). What is the problem with this approach (most complete and precise answer counts)
a. This is not how psychology works
b. The k‐NN algorithm relies on distance, so normalizing the training set should be done for higher accuracy of the similarity estimate
c. The k‐NN algorithm is sensitive to the local structure of the data
d. The k‐NN algorithm represents feature matching in Euclidian space. The weights are spatial, not cognitive or affective
Antwoord: a. True but imprecise
b. Perhaps but the question is about the problem with distance weights
c. True but irrelevant to the question
d. RIGHT ANSWER (complete description is in the Example exam questions and answers file