Title :
Power Law In Feedforward Networks
Author :
Câteau, Hideyuki
Author_Institution :
University of Tokyo
Abstract :
When a feedforward network memorizes several different iterns, the pace of the memory declines due to an interference between already memorized items. We show that this slow down of the learning is generally described by a power law.
Keywords :
Cost function; Feeds; Humans; Intelligent networks; Interference; Neural networks; Neurons; Physics; Psychology; Vectors;
Conference_Titel :
Neural Networks, 1993. IJCNN '93-Nagoya. Proceedings of 1993 International Joint Conference on
Print_ISBN :
0-7803-1421-2
DOI :
10.1109/IJCNN.1993.713966
