CMPUT 261, Fall 2025
Practice Midterm Exam
Total points: 49
This exam has 6 pages
This exam is CLOSED BOOK. You are not allowed to use any resources or equip-
ment other than this exam, a pen/pencil, a non-programmable calculator, and a
single-page “cheat sheet”.
You are not allowed to use your own scratch paper. Instead, use backsides of the
exam pages as scratch paper for your work.
You may use a non-programmable calculator if you wish.
You are not allowed to use any other electronic equipment (e.g., phones, computers,
tablets, etc.).
You are not allowed to communicate with anybody except the examiners/proctors.
Name:
CCID: @ualberta.ca
, CMPUT 261 Practice Midterm — CCID: 2
1. (Short Answers; 16 points)
(a) [2 points] What is an admissible heuristic?
(b) [2 points] Is breadth first search complete? Why or why not?
(c) [2 points] Explain why variable elimination can be more efficient than just computing
the full joint distribution and then normalizing.
(d) [2 points] Which of breadth-first search, least-cost-first search, and A∗ search has the
best worst-case time complexity? Why?
(e) [2 points] Why is loss on the training dataset not a good estimate for generalization
performance?
(f) [2 points] What is a loss function?
(g) [2 points] Give an expression for the squared error loss L(θ) for a hypothesis h(x; θ)
with parameters θ on a dataset with two examples: {(x1 , y1 ), (x2 , y2 )}
Practice Midterm Exam
Total points: 49
This exam has 6 pages
This exam is CLOSED BOOK. You are not allowed to use any resources or equip-
ment other than this exam, a pen/pencil, a non-programmable calculator, and a
single-page “cheat sheet”.
You are not allowed to use your own scratch paper. Instead, use backsides of the
exam pages as scratch paper for your work.
You may use a non-programmable calculator if you wish.
You are not allowed to use any other electronic equipment (e.g., phones, computers,
tablets, etc.).
You are not allowed to communicate with anybody except the examiners/proctors.
Name:
CCID: @ualberta.ca
, CMPUT 261 Practice Midterm — CCID: 2
1. (Short Answers; 16 points)
(a) [2 points] What is an admissible heuristic?
(b) [2 points] Is breadth first search complete? Why or why not?
(c) [2 points] Explain why variable elimination can be more efficient than just computing
the full joint distribution and then normalizing.
(d) [2 points] Which of breadth-first search, least-cost-first search, and A∗ search has the
best worst-case time complexity? Why?
(e) [2 points] Why is loss on the training dataset not a good estimate for generalization
performance?
(f) [2 points] What is a loss function?
(g) [2 points] Give an expression for the squared error loss L(θ) for a hypothesis h(x; θ)
with parameters θ on a dataset with two examples: {(x1 , y1 ), (x2 , y2 )}
