Abstract
Multiple novelty input neural networks have been presented. It is shown that this method produces relatively good recognition accuracy for an unconstrained handwritten numeral database.
Original language | English |
---|---|
Pages (from-to) | 1112-1113 |
Number of pages | 2 |
Journal | Electronics Letters |
Volume | 34 |
Issue number | 11 |
DOIs | |
State | Published - 28 May 1998 |