Incremental learning in autonomous systems: evolving connectionist systems for on-line image and speech recognition

The paper presents an integrated approach to incremental learning in autonomous systems, that includes both pattern recognition and feature selection. The approach utilizes evolving connectionist systems (ECoS) and is applied on on-line image and speech pattern learning and recognition tasks. The experiments show that ECoS are a suitable paradigm for building autonomous systems for learning and navigation in a new environment using both image and speech modalities.