Title :
Second-order generalization
Author :
Neville, Richard S.
Author_Institution :
Dept. of Comput., UMIST, Manchester, UK
Abstract :
Second-order generalization is usually of a more abstract nature than standard generalization, as unseen stimuli may be classified by some higher order rule. The research presents a new theory, which utilises information inheritance in order to perform second-order generalization. In this article, we present the theory behind a recognition system, which performs recognition without learning. Recognition without learning can be viewed as one form of second-order generalization.
Keywords :
generalisation (artificial intelligence); neural nets; pattern classification; information inheritance; neural networks; recognition system; second order generalization; Electronic mail; Humans; Image recognition; Mirrors; Network topology; Neural networks; Neurons; Optimization methods; System testing;
Conference_Titel :
Neural Networks, 2004. Proceedings. 2004 IEEE International Joint Conference on
Print_ISBN :
0-7803-8359-1
DOI :
10.1109/IJCNN.2004.1380060