Final answer:
In a Bayesian network, a Boolean variable with k Boolean parents requires 2^k - 1 rows in its Conditional Probability Table (CPT) to account for all combinations of parent states, with one state's probability implied by normalization.
Step-by-step explanation:
To determine the number of rows needed to explicitly store the Conditional Probability Table (CPT) for a Boolean variable Xi with k Boolean parents in a Bayesian network (Bnet), we need to consider all possible combinations of truth values for the parents. Since each of the k parents can be either true or false, there are 2k possible combinations of parent states. However, since the variable itself is also Boolean, we can represent one of its states (true or false) using the normalization condition of probabilities (that is, the probabilities sum to 1), which means we only need to store explicit values for one less than the total number of combinations. Therefore, for a CPT of a Boolean variable Xi, we need to explicitly store 2k - 1 rows to account for all the potential combinations of its parents' states while leaving one state of the variable itself to be implicitly defined.