scientific article; zbMATH DE number 4087481

From MaRDI portal
Publication:3815906





[For the entire collection see Zbl 0661.00018.] Neural networks, as found in the nervous system, possess impressive computational powers which are attributed largely to their high (feedback) interconnectivity. We investigate the theoretical foundation of the architecture of neural networks in order to design and systematically specify their functional organization. The objective is to establish a sound framework for design which would substantiate the computational properties attributed to neural nets. Our emphasis throughout will be on the design of associative memory or content- addressable memory (CAM). This framework takes into account the characteristics and the limitations of the physical components which are the building blocks for the proposed associative devices. We also present conditions that given vectors must satisfy in order to successfully store them in the network. Our conditions suggest a preprocessing procedure of the desired memories to render them storable.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3815906)