论文部分内容阅读
极限学习机(ELM)是一种新型的单隐含层神经网络的训练方法,同传统的基于梯度的网络训练方法相比,具有快速的学习速度和更好的泛化性能.ELM在实际应用中往往需要大量的隐含层神经元,由于随机设定输入权值和偏置值,容易导致病态问题的出现.为解决上述问题,提出一种应用量子粒子群(QPSO)优化包括隐含层节点个数在内的网络参数的方法.这种优化基于验证集的均方根误差,考虑到了输入权值矩阵的范数.在典型的回归和分类问题上进行试验证明了算法的有效性.“,”Extreme Learning Machine (ELM) is a novel learning algorithm for Single-Hidden-Layer Feed Forward Neural Networks (SLFN) with much faster learning speed and better generalization performance than traditional gradient-based learning algorithms.However ELM tends to require more neurons in the hidden layer and lead to ill-conditioned problem due to the random selection of input weights and hidden biases.To address these problems,a learning algorithm was proposed which used quantum-behaved particle swarm optimization (QPSO) to select the optimal network parameters including the number of hidden layer neurons according to the both the root mean square error on validation data set and the norm of output weights.Experimental results on benchmark regression and classification problems have verified the performance and effectiveness of the proposed approach.