Question 15 (1 point)

The perceptron algorithm is guaranteed to converge to a solution when

Question 15 options:

	the attributes are binary
	the data is linearly separable
	there are only two classes
	all of the above
	none of the above

Question 16 (1 point)

The iterations in the perceptron algorithm where you consider changing the weight vector based on a dot product with some instance vector end when

Question 16 options:

	you reach the last instance
	the weight vector has not changed for all of the instances
	possibly never (unless there is a numerical limit on the number of iterations)
	none of the above
	more than one of the above

Question 17 (1 point)

Use the perceptron algorithm to find the hyperplane separating the two classes. The data are (a1, a2, class): {(1, 4, 1), (-1, -1, -1)}. What are the final weights for a1 and a2?

Question 17 options:

	w1=1, w2=1
	w1=1, w2=4
	w1=0, w2=0
	w1=-1, w2=-4
	none of the above

Question 18 (1 point)

For the dataset in Table 1, using the PRISM algorithm to find rules for the class mammal, the first rule generated would be:

Table 1: Vertebrate Dataset (modified from Introduction to Data Mining by Tan et al.)

Name	Body temp	Give birth	Aquatic	Class
human	warm	yes	no	mammal
echidna	warm	no	no	mammal
salmon	cold	no	yes	fish
whale	warm	yes	yes	mammal
eel	cold	no	yes	fish
bat	warm	yes	no	mammal
shark	cold	yes	yes	fish
cat	warm	yes	no	mammal

Question 18 options:

	if Body temp = warm, then class = mammal
	if Give birth = yes, then class = mammal
	if Aquatic = no, then class = mammal
	a or c
	none of the above