论文部分内容阅读
在多个单分类器的基础上进行集成,能有效提高系统的识别能力。它已成为模式识别的重要方法。控制系统告诉我们,反馈的引入能大大提高系统的性能。因此,本文提出反馈集成网络模型。由于所提模型是复杂的动力学系统,我们简要介绍了有关的数字结论,然后详细给出它的监督学习算法。实验表明,我们的模型比普通前向型人工神经网络更为优越。
The integration based on multiple single classifiers can effectively improve the system’s recognition ability. It has become an important method of pattern recognition. The control system tells us that the introduction of feedback can greatly improve the performance of the system. Therefore, this paper proposes a feedback integrated network model. Since the proposed model is a complex dynamical system, we briefly introduce the relevant numerical conclusions and then give details of its supervised learning algorithm. Experiments show that our model is superior to ordinary forward-type artificial neural network.