Question

Which of the following statements is/are correct in a Bayesian network?

• A. Variable elimination is an approximate inference algorithm
• B. Gibbs sampling is an exact inference algorithm
• C. Variable elimination is used to determine conditional probabilities
• D. Rejection sampling is an approximate inference algorithm

Hint:

In Bayesian networks, inference algorithms can be exact or approximate. Exact algorithms like variable elimination are computationally expensive, while approximate methods like Gibbs sampling and rejection sampling provide feasible solutions for large networks.

Answer 1

The Correct Option is C, D

Let's evaluate each option:

Option (A): Variable elimination is an exact inference algorithm, not an approximate one. It computes the exact marginal distributions but may become computationally expensive for large networks. Hence, this statement is incorrect.
Option (B): Gibbs sampling is an approximate inference algorithm. It uses sampling to estimate the distribution, which is approximate. Therefore, this option is incorrect.
Option (C): Variable elimination is indeed used to compute conditional probabilities in Bayesian networks. It works by eliminating variables one at a time to compute the marginal distribution or conditional probabilities. Thus, this option is correct.
Option (D): Rejection sampling is an approximate inference algorithm. It involves generating random samples and rejecting those that do not meet the criteria. It is used when exact inference is difficult, making this option correct.

Thus, the correct answer is (C) and (D).