E.: Evolutionary concept learning

Inductive learning in First-Order Logic (FOL) is a hard task due to both the pro-hibitive size of the search space and the computational cost of evaluating hypotheses. This paper introduces an evolutionary algo-rithm for concept learning in (a fragment of) FOL. The algorithm evolves a population of Horn clauses by repeated selection, mutation and optimization of more fit clauses. Its main novelty, with respect to previous approaches, is the use of stochastic search biases for re-ducing the complexity of the search process and of the clause fitness evaluation. An ex-perimental evaluation of the algorithm indi-cates its effectiveness in learning short hy-potheses of satisfactory accuracy in a short amount of time.