竞争监督学习法在集成型识别系统中的应用
Competitive Supervised Learning Algorithm and Application in Integrated Recognition System
-
摘要: 在集成方法中,神经网络集成方法对研制集成型模式识别系统是有效的.但是,单个 子分类器和集成网络的性能对集成系统的整体识别效果都有影响.因此,要进一步提高系统 性能也必须改进子分类器和集成网络.文中采用竞争监督学习法,构造一个网络集成系统,用 于手写数字字符识别.实验证明,该方法的确能够改进系统的收敛速度和泛化能力.Abstract: In pattern recognition application,integration of artificial neural networks has been recognized as a powerful tool,but the performance of the sub-classifiers has effect on the performance of the whole integration system. In this paper,we apply the proposed aIgorithm--Competitive Supervised Learning Algorithm to the integrated recognition neural network system.Experimental results have confirmed that the proposed algorithm improves learning and generalization of the system.
-
Key words:
- Integration /
- competitive supervised learning
-
计量
- 文章访问数: 2063
- HTML全文浏览量: 118
- PDF下载量: 1062
- 被引次数: 0
下载:
