Title :
Functional representation of rescaling process and memory capacity
Author_Institution :
Dept. of Comput. Sci. & Syst. Eng., Muroran Inst. of Technol., Hokkaido, Japan
Abstract :
The data retrieval process in hetero-associative memory, extended to higher order correlations, has been formulated into a functional in terms of Hamming distance between the memorized keys and an applied input key. In this paradigm, we show that the functional representation already presented is exactly the same as the functional representation in terms of the Krawtchouk polynomial, and further, the memory capacity of associative memory is also described by a functional form in terms of the Krawtchouk polynomial. The resultant functional representation on memory capacity is of wide generality and it is verified that it includes the well-known formula on memory capacity of the Hopfield auto-associative memory as a special case
Keywords :
Hopfield neural nets; content-addressable storage; polynomials; Hamming distance; Hopfield auto-associative memory; Krawtchouk polynomial; applied input key; associative memory; data retrieval process; functional form; functional representation; hetero-associative memory; higher order correlations; memorized keys; memory capacity; rescaling process; Associative memory; Computer science; Data engineering; Hamming distance; Information retrieval; Neurons; Polynomials; Prototypes; Systems engineering and theory;
Conference_Titel :
Neural Networks, 2000. IJCNN 2000, Proceedings of the IEEE-INNS-ENNS International Joint Conference on
Conference_Location :
Como
Print_ISBN :
0-7695-0619-4
DOI :
10.1109/IJCNN.2000.861539