1、Semi-Supervised and Unsupervised Extreme Learning Machines
Gao Huang, Shiji Song, Jatinder N. D. Gupta, and Cheng Wu
ELM与manifold regularization结合,完成semi-supervised ELM (SS-ELM) and the unsupervised ELM (US-ELM)
ELM:
输出权重objective function:minimizing the sum of the squared losses of the prediction error
求解优化函数:ridge regression or regularized least squares
condition1:the number of training patterns is larger than the number of the hidden neurons, the above equation is overdetermined
condition2: the number of training patterns is smaller than the number of hidden neurons
restrict β to be a linear combination of the rows of H: β = HTα
manifold regularization:
参考降维方法:Laplacian Eigenmaps (拉普拉斯特征映射)——拉普拉斯矩阵