IBL & NN

Question 1

What is a Nearest Neighbour?

Answer 1

The closest data point: maximum similarity or minimum distance.

Question 2

Nearest Neighbour variants

Answer 2

1. 1NN

2. KNN

Question 3

1NN

Answer 3

Classify the test input according to the class of the
closest training instance

Question 4

KNN

Answer 4

Classify the test input according to the majority class
of the K nearest training instances

Question 5

weighted K-NN

Answer 5

Classify the test input according to the weighted accumulative class of the K nearest training instances, where weights are based on similarity of the input to each of the K neighbours

Question 6

offset-weighted K-NN

Answer 6

Classify the test input according to
the weighted accumulative class of the K nearest training instances, where weights are based on similarity of the input to each of the K neighbours, factoring in an offset for the prior expectation of a test input being a member of that class