Natural language grammatical inference with recurrent neural networks

This paper examines the inductive inference of a complex grammar with neura l networks-specifically, the task considered is that of training a network to classify natural language sentences as grammatical or ungrammatical. the reby exhibiting the same kind of discriminatory power provided by the Princ iples and Parameters linguistic framework, or Government-and-Binding theory . Neural networks are trained, without the division into learned vs. innate components assumed by Chomsky, in an attempt to produce the same judgments as native speakers on sharply grammatical/ungrammatical data. How a recurr ent neural network could possess linguistic capability and the properties o f various common recurrent neural network architectures are discussed. The problem exhibits training behavior which is often not present with smaller grammars and training was initially difficult. However, after implementing several techniques aimed at improving the convergence df the gradient desce nt backpropagation-through-time training algorithm, significant learning wa s possible. It was found that certain architectures are better able to lear n an appropriate grammar. The operation of the networks and their training is analyzed. Finally, the extraction of rules in the form of deterministic finite state automata is investigated.