基于快速组合神经网络分类器的数字调制识别新方法

Research and Application on an Effective Combination Classifier of Neural Network

单个神经网络分类器无法提供一种对所有数据进行精确解释的能力，但是会对部分数据贡献一些有用的信息，组合神经网络分类器利用有效的组合算法集成单个分类器，使分类能力得到显著提高。本文提出一种快速非线性组合算法对通信信号的调制类型进行分类，取得了较好的效果，实验结果证明了此方法的有效性。

No single neural network provides the ability to precisely interpret the data. Rather, a number of neural networks each contribute some useful cues. The combination classifier of neural network integrates single by effective combination algorithm, so that ability classification is improved notably. This paper suggests a non-linear combination algorithm to classify modulation types of communication signal. its result is inspiring. Test has proved its good performance.