CAJ | 학술논문

针对传统的批量学习算法学习速度慢、对空间需求量高的缺点，提出了一种基于簇的极限学习机的在线学习算法。该算法将分簇的理念融入到极限学习机中，并结合极限学习机，提出了一种基于样本类别和样本输出的分簇标准；同时提出了一种加权的Moore-Penrose算法求隐层节点与输出节点的连接权重。实验结果表明，该算法具有学习能力好、拟合度高、泛化性能好等优点。
침대전통적비량학습산법학습속도만、대공간수구량고적결점，제출료일충기우족적겁한학습궤적재선학습산법。해산법장분족적이념융입도겁한학습궤중，병결합겁한학습궤，제출료일충기우양본유별화양본수출적분족표준；동시제출료일충가권적Moore-Penrose산법구은층절점여수출절점적련접권중。실험결과표명，해산법구유학습능력호、의합도고、범화성능호등우점。
Traditional batch learning algorithm is slow to learn and has a high demand for space. This paper proposes a clutering_based and ELM_based online learning algorithm. In this algorithm, it takes the concept of clustering into extreme learning machine, combines with extreme learning machine, proposes a category_based, output_based standard of cluster-ing. At the same time, it also proposes a weighted Moore-Penrose algorithm to solve the weight vector connecting the hidden nodes and the output nodes. The result shows that this algorithm has good learning ability and high goodness of fit, pro-duces better generalization performance, and so on.