CS7646 Week 2 Multi-Select Practice Exam
UPDATED QUESTIONS AND CORRECT
ANSWERS
Q1 — Epistemic uncertainty
A. Epistemic uncertainty arises from limited knowledge or limited data.
B. Epistemic uncertainty disappears once a model is trained.
C. Additional data can reduce epistemic uncertainty.
D. Epistemic uncertainty reflects randomness inherent in the data generating process.
E. Epistemic uncertainty is sometimes called model uncertainty. - CORRECT
ANSWER (T, F, T, F, T)
Q2 — Aleatoric uncertainty
A. Aleatoric uncertainty reflects randomness in the real world.
B. Aleatoric uncertainty can be eliminated by collecting more data.
C. Aleatoric uncertainty is often described as noise.
D. Aleatoric uncertainty exists even with perfect models.
E. Aleatoric uncertainty occurs only in classification problems. - CORRECT
ANSWER (T, F, T, T, F)
Q3 — Probability interpretations
A. The frequentist view interprets probability as long-run frequency of events.
B. Bayesian probability measures degree of belief about an outcome.
C. Bayesian probability cannot incorporate prior information.
D. Frequentist probability typically requires repeated experiments.
E. Bayesian probability cannot be used in machine learning. - CORRECT
ANSWER (T, T, F, T, F)
Q4 — Joint probability
, A. Joint probability measures the probability of two events occurring together.
B. Joint probabilities must always sum to more than one.
C. Joint probability distributions can describe relationships between variables.
D. Joint probabilities can be used to compute marginal probabilities.
E. Joint probability is unrelated to conditional probability. - CORRECT ANSWER (T,
F, T, T, F)
Q5 — Marginal probability
A. Marginal probability considers one variable independent of others.
B. Marginal probabilities can be obtained by summing joint probabilities.
C. Marginalization ignores the influence of other variables.
D. Marginal probability requires independence between variables.
E. Marginal probabilities are useful for simplifying probability models. - CORRECT
ANSWER (T, T, T, F, T)
Q6 — Conditional probability
A. Conditional probability measures probability given additional information.
B. Conditional probabilities always sum to zero.
C. Conditional probability allows probability to depend on known conditions.
D. Conditional probabilities can be used to model relationships between variables.
E. Conditional probability is unrelated to joint probability. - CORRECT ANSWER (T,
F, T, T, F)
Q7 — Bayes' theorem intuition
A. Bayes' theorem allows reversing conditional probabilities.
B. Bayes' theorem updates beliefs after observing new data.
C. Bayes' theorem eliminates uncertainty completely.
D. Bayes' theorem combines prior beliefs with observed evidence.
E. Bayes' theorem is only applicable in medical diagnosis problems. - CORRECT
ANSWER (T, T, F, T, F)
UPDATED QUESTIONS AND CORRECT
ANSWERS
Q1 — Epistemic uncertainty
A. Epistemic uncertainty arises from limited knowledge or limited data.
B. Epistemic uncertainty disappears once a model is trained.
C. Additional data can reduce epistemic uncertainty.
D. Epistemic uncertainty reflects randomness inherent in the data generating process.
E. Epistemic uncertainty is sometimes called model uncertainty. - CORRECT
ANSWER (T, F, T, F, T)
Q2 — Aleatoric uncertainty
A. Aleatoric uncertainty reflects randomness in the real world.
B. Aleatoric uncertainty can be eliminated by collecting more data.
C. Aleatoric uncertainty is often described as noise.
D. Aleatoric uncertainty exists even with perfect models.
E. Aleatoric uncertainty occurs only in classification problems. - CORRECT
ANSWER (T, F, T, T, F)
Q3 — Probability interpretations
A. The frequentist view interprets probability as long-run frequency of events.
B. Bayesian probability measures degree of belief about an outcome.
C. Bayesian probability cannot incorporate prior information.
D. Frequentist probability typically requires repeated experiments.
E. Bayesian probability cannot be used in machine learning. - CORRECT
ANSWER (T, T, F, T, F)
Q4 — Joint probability
, A. Joint probability measures the probability of two events occurring together.
B. Joint probabilities must always sum to more than one.
C. Joint probability distributions can describe relationships between variables.
D. Joint probabilities can be used to compute marginal probabilities.
E. Joint probability is unrelated to conditional probability. - CORRECT ANSWER (T,
F, T, T, F)
Q5 — Marginal probability
A. Marginal probability considers one variable independent of others.
B. Marginal probabilities can be obtained by summing joint probabilities.
C. Marginalization ignores the influence of other variables.
D. Marginal probability requires independence between variables.
E. Marginal probabilities are useful for simplifying probability models. - CORRECT
ANSWER (T, T, T, F, T)
Q6 — Conditional probability
A. Conditional probability measures probability given additional information.
B. Conditional probabilities always sum to zero.
C. Conditional probability allows probability to depend on known conditions.
D. Conditional probabilities can be used to model relationships between variables.
E. Conditional probability is unrelated to joint probability. - CORRECT ANSWER (T,
F, T, T, F)
Q7 — Bayes' theorem intuition
A. Bayes' theorem allows reversing conditional probabilities.
B. Bayes' theorem updates beliefs after observing new data.
C. Bayes' theorem eliminates uncertainty completely.
D. Bayes' theorem combines prior beliefs with observed evidence.
E. Bayes' theorem is only applicable in medical diagnosis problems. - CORRECT
ANSWER (T, T, F, T, F)
