How does best first search generate feature subset in case of correlation based feature selection (CFS)?

Question

I am trying to get important subset of features using CFS. I understood most of the part of the algorithm, stuck on the part where feature subsets are generated for evaluation. There are different search algorithms, best first search (BFS) is mostly used. I want to know how does BFS make different combination of subsets of feature?

Answer 1

等待大神解答