基于多层激励函数量子神经网络的字符识别算法  

Character Recognition Algorithm Based on Multi-Level Transfer Function Quantum Neural Network

作　　者：吴茹石[1] 朱大奇[2] 彭力[1] 

机构地区：[1]江南大学智能控制研究所,无锡214122 [2]上海海事大学信息工程学院,上海200135

出　　处：《数据采集与处理》2007年第4期401-406,共6页Journal of Data Acquisition and Processing

基　　金：江苏省自然科学基金(BK2004021)资助项目;教育部科学技术研究重点项目基金(105088)资助项目

摘　　要：针对不同样本之间存在交叉数据的模式识别问题,将多层激励函数的量子神经网络引入模式识别之中,提出一种基于量子神经网络的模式识别算法。量子神经网络是将神经元与模糊理论相结合的模糊神经系统,由于自身固有的模糊性,它能将决策的不确定性数据合理地分配到各模式中,从而减少模式识别的不确定度,提高模式识别的准确性。本文以英文字母为例,应用量子神经网络模型进行字符识别,通过比较发现量子神经网络除了可以克服BP网络的诸多缺点外,对具有不确定性、两类模式之间存在交叉数据的模式识别问题,有极好的分类效果。仿真结果证明该方法的正确性和有效性。Aiming at the data overlapping of different patterns on pattern recognition, a pattern recognition algorithm is presented based on the multi-level transfer function quantum neural network （QNN）. QNN is a kind of neuro-fuzzy system by merging neural modeling with fuzzytheoretic concepts, because of the inherent ambiguity it can assign the ambiguity data to the corresponding patterns reasonably, thus reducing the uncertainty of pattern recognition and improving the veracity of pattern recognition. An example for English character recognition is provided and compared with the BP network. Results show that the QNN network can overcome BP network limitations, and has perfect classification effect for pattern recognition problems which have uncertainty and overlapping data between two patterns. The validity and the effectiveness are demonstrated by simulation.

关 键 词：字符识别算法 量子神经网络 多层激励函数 模式识别 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]