Abstract

Current knowledge from cognitive sciences impacts also modern information technologies and their application in practice, e.g., for natural language processing or computer vision. New perspectives are thus opening in the areas of structural data analysis, restarting automata (RA) and artificial neural networks. The project is expected to bring new methods for structural data analysis and design of 2D-restarting automata (RA). The methods will use training (of neural networks and RA) and support the recognition of natural hierarchies in the data (structure) and its modeling. We plan to test them in 2 pilot studies.

Our goals are

• Design new methods for efficient (structural) data analysis. Extend these methods to extract also the information concerning (hierarchical) structure of the processed data, its development and impact to recall performance.
• Extend the original model of RA with means for grammatical inference over multi-dimensional inputs, e.g., pictures. Analyze theoretical properties of the proposed 2D-RA, and 2D-grammars, resp.
• Implement the developed methods and test them with the aim to assess their limits in practical applications. Use the developed software modules in 2 pilot studies – for (structural) data analysis and in image recognition.