EELM trains a single hidden layer forward neural network with Sigmoidal function as activation function
Keeps the column rank of the hidden layer output matrix full