为了减小传统的极限学习机网络的规模及提高网络的泛化性能,利用Akaike信息准则作为学习的最优停止准则以选择合适的隐层节点数量,同时利用修正Gram-Schmidt算法自动调整网络参数,提出改进的极限学习机网络构造算法。通过与传统极限学习机在通用标杆问题上的实验结果比较表明,该改进的极限学习机具有更精简的网络结构和更快的学习速度,同时具有良好的学习精度。 In order to reduce the size of the traditional ELM network and improve the generalization of the network, Akaike information criterion is used as the optimal stopping criterion to select the appropriate number of hidden nodes. At the same time, the modified Gram-Schmidt algorithm is used to automatically adjust the network parameters , Proposed an improved limit learning machine network construction algorithm. Compared with the traditional extreme learning machine on the general benchmark problem, the experimental results show that the improved extreme learning machine has a more streamlined network structure and faster learning speed, and has good learning accuracy.