CS 7641 Machine Learning Unit 2 Exam Practice Questions ( Verified UPDTE!!!!!).pdf

Page |1


CS 7641 Machine Learning Unit 2 Exam Practice Questions
(2026-2027 Verified UPDTE!!!!!)



Instructions: This exam consists of multiple-choice questions.
Choose the best answer for each question. The questions are
organized by the major topic areas within Unit 2: Randomized
Optimization, Clustering, Dimensionality Reduction & Feature
Selection, and Theoretical Concepts.


Section 1: Randomized Optimization (Questions)


1. What are the four optimization approaches covered in CS
7641?
A) Gradient descent, Newton's method, conjugate gradient,
BFGS
B) Generate and test, calculus, Newton's Method, Randomized
Optimization
C) Linear programming, quadratic programming, convex
optimization, stochastic optimization
D) Supervised, unsupervised, reinforcement, semi-supervised
Answer: B
Rationale: The four optimization approaches are: 1) Generate

, Page |2


and test, 2) Calculus, 3) Newton's Method, and 4) Randomized
Optimization.
2. What is the hill climbing algorithm?
A) A global optimization algorithm that always finds the optimal
solution
B) A local search algorithm that iteratively moves to the best
neighboring solution
C) A clustering algorithm
D) A supervised learning algorithm
Answer: B
Rationale: Hill climbing is a local search algorithm that starts
with an initial solution and iteratively moves to the best
neighboring solution, stopping when no improvement is found.
3. What is a key limitation of hill climbing?
A) It always finds the global optimum
B) It can get stuck in local optima
C) It is too slow
D) It requires labeled data
Answer: B
Rationale: Hill climbing can get stuck in local optima because it
only moves to neighbors that improve the objective, unable to
escape local peaks.
4. What is randomized hill climbing?
A) A variant of hill climbing that randomly samples neighbors

, Page |3


instead of exhaustively checking all neighbors
B) A deterministic local search
C) A clustering algorithm
D) A supervised learning algorithm
Answer: A
Rationale: Randomized hill climbing samples neighbors
randomly rather than evaluating all possible neighbors, which
can be more efficient in large search spaces.
5. What is simulated annealing?
A) A deterministic local search
B) A randomized optimization algorithm that allows uphill
moves with a probability that decreases over time
C) A clustering algorithm
D) A supervised learning algorithm
Answer: B
Rationale: Simulated annealing allows uphill moves (worse
solutions) with a probability that decreases according to a
cooling schedule, helping escape local optima.
6. What is the Metropolis rule in simulated annealing?
A) An uphill move with cost increase ΔE > 0 at temperature T is
accepted with probability exp(-ΔE/T)
B) All moves are accepted
C) No moves are accepted
D) Only downhill moves are accepted

, Page |4


Answer: A
Rationale: The Metropolis rule accepts an uphill move with cost
increase ΔE at temperature T with probability exp(-ΔE/T). A
downhill move (ΔE < 0) is accepted with probability 1.
7. What is the role of temperature in simulated annealing?
A) To control the learning rate
B) To control the probability of accepting worse solutions;
higher temperature means more exploration
C) To control the number of iterations
D) To control the exploration rate
Answer: B
Rationale: Temperature T controls the probability of accepting
worse solutions. Higher temperatures allow more exploration;
as T decreases, the algorithm becomes more greedy.
8. What is geometric cooling in simulated annealing?
A) T_k = T_0 · α^k, where α is the cooling rate
B) T_k = T_0 - k
C) T_k = T_0 / k
D) T_k = T_0 + k
Answer: A
Rationale: Geometric cooling uses T_k = T_0 · α^k, where T_0 is
the initial temperature and α is the cooling rate (typically
between 0 and 1).