Children as Models for Computers: Natural Language Acquisition for Machine Learning

Abstract

Abstract. This paper focuses on a subfield of machine learning, the socalled grammatical inference. Roughly speaking, grammatical inference deals with the problem of inferring a grammar that generates a given set of sample sentences in some manner that is supposed to be realized by some inference algorithm. We discuss how the analysis and formalization of the main features of the process of human natural language acquisition may improve results in the area of grammatical inference.