Enhanced Extreme Learning Machine with Modified Gram-Schmidt Algorithm

Extreme learning machine (ELM) has shown to be extremely fast with better generalization performance. However, the implementation of ELM encounters two problems. First; ELM tends to require more hidden nodes than conventional tuning-based algorithms. Second, subjectivity is involved ill choosing hidden nodes number. In this paper, we apply the modified Gram-Schmidt (MGS) method to select; hidden nodes which maximize the increment to explained variance of the desired output. The Akaike's final prediction error (FEE) criterion are used to automatically determine the number of hidden nodes. In comparison with conventional ELM learning method on several commonly used regressor benchmark problems. our proposed algorithm can achieve compact network with much faster response and satisfactory accuracy.