与传统的前馈神经网络不同，新加坡南洋理工大学教授黄广斌等人[8]提出的

Unlike traditional feed-forward neural networks, the extreme learning machine ELM learning speed proposed by Huang Guangbin, a professor at Nanyang Technological University in Singapore [8] has greatly improved compared with traditional learning algorithms, and has better generalization ability. It is proved that the algorithm can approximate the training sample with zero error when the number of hidden layer neurons is equal to the number of training samples, and when the number of hidden neurons is less than the number of training samples, the training error of ELM can approximate an arbitrary value greater than 0.

Unlike traditional forward-feed neural networks, Professor Huang Guangbin of Nanyang Technological University in Singapore and others proposed a significant improvement of the extreme learning machine ELM learning speed than the traditional learning algorithm, and has a better generalization ability, by proving that the algorithm in the implied layer of neurons equal to the number of training samples can be zero error approximation of training samples, and when the number of hidden neurons is less than the number of training samples, THE ELM training error can approach a value greater than any.

Different from the traditional feedforward neural network, the limit learning machine (ELM) proposed by Huang guangbin, Professor of Nanyang University of technology, Singapore Compared with the traditional learning algorithm, the learning speed is greatly improved, and it has better generalization ability. It is proved that the algorithm can approach the training sample with zero error when the number of neurons in the hidden layer is equal to the number of training samples, and the training error of elm can approach any value greater than 0 when the number of hidden neurons is less than the number of training samples.<br>