Development of autonomic neural board for banknotes and advancement to palm prints recognition
Fumiaki Takeda, Toshihiro Nishikage, Yoshiyuki Matsumoto · 2003
The authors (1996, 1998) have previously proposed a banknote recognition system using a neural network to develop new types of banking machines such as banknote readers and sorters. In this neural recognition system, the banknotes data have to be transported from the banking machines to a personal computer (PC) for learning. After learning on the PC, the neural parameters such as weights and others are then downloaded to the neural board. However, one has to cope with the recovery of the recognition ability for various fluctuations of the banknotes in the field. In this paper, we implement the neural learning algorithm to this board enabling it to execute learning by itself and extend its specification to realize the intelligent machines. We construct the experimental system. Then we show the effectiveness and possibility of robust autonomic learning using banknotes and palm prints data.