A Comparative Study On The Find-S And Candidate Elimination Algorithms

The central idea of inductive learning is algorithms of concept learning. Concept leaning is to induce the general rules from specific training examples and the rules are expected to both cover the positive examples and exclude the negative examples. Concept learning can be formulated as a problem of searching through a predefined space of potential hypotheses for the hypothesis that best fits the training examples. First of all, this paper briefly presents some basic knowledge in machine learning and concept learning, and then makes a comparative research on the Find-S and candidate elimination algorithms, both of which are the most representative algorithms for concept learning. The two algorithms for concept learning efficiently complete the whole search through the hypothesis space based on a very useful structure: the general-to-specific partial order. This partial order structure can be extended to any concept learning problem, and therefore, it plays an instructive role in all algorithms for concept learning.The comparative study in this paper shows the advantages and disadvantages of the two algorithms. It aims to provide some useful guidelines for selecting an appropriate one from the two algorithms when they are applied to real problems.