ELM算法指出，其实隐层的权值矩阵W和偏置b其实是没有必要调整的，在学习算法开始时，任意随机给定W和b的值，利用其计算出H（隐层节点的输出），并令其保持不变，需要确定的参数就只有β了。这是一个比较重要的理论基础。(The ELM algorithm is pointed out, in fact, hidden layer weights matrix W and B is not necessary to adjust the bias of the learning algorithm, in the beginning, randomly given W and b values, the calculated H (output nodes), and remained unchanged, only need to determine the parameters of the beta. This is a very important theoretical basis.)