论文部分内容阅读
利用聚类对噪声干扰的抵抗能力和对好的样本进行强化学习的思想，引入了聚类分析和鲁棒因子，提出一种新颖的鲁棒学习算法（包括了选择不同鲁棒因子而构成的鲁棒算法１ 和鲁棒算法２），并对三维曲面和混合噪音进行了仿真实验研究。仿真结果表明，该算法在鲁棒性、收敛性方面明显优于普通的ＢＰ算法。
This paper introduces clustering analysis and robust factor by using clustering’s resistance to noise interference and enhancement learning of good samples. A novel robust learning algorithm is proposed (including the selection of different robust factors Robust Algorithm 1 and Robust Algorithm 2), and simulated the 3D surface and mixed noise. The simulation results show that the proposed algorithm is superior to ordinary BP algorithm in robustness and convergence.