Higher-Order Temporal Generalization

The research will investigate different types of recurrent artificial neural networks and their capabilities with respect to higher-order temporal generalization. This research will initially utilize a particular type of recurrent net the Elman’s topology for its initial experimental work; it will then move on to more complex recurrent nets. Higher-order temporal generalization methodologies will be investigated. The investigations will develop different methodologies, be they: mathematical, algorithmic, pre- and post-training techniques for enhancing the higher-order temporal generalization ability of artificial neural networks. In subsequent research the researcher will investigate advanced methodologies that will enable artificial neural networks to perform higher-order temporal generalization more efficiently.