Global convergence of Negative Correlation Extreme Learning Machine

Ensemble approaches introduced in the Extreme Learning Machine literature mainly come from methods that rely on data sampling procedures, under assumption training are heterogeneously enough to set up diverse base learners. To overcome this assumption, it was proposed an ELM ensemble method based Negative Correlation framework, called (NCELM). This model works two stages: (i) different ELMs generated as learners with random weights hidden layer, and (ii) a NCL penalty term information of prediction is each minimization problem, updating learners, (iii) second step iterated until converges. Although validated by experimental study multiple benchmark datasets, no given conditions about convergence. paper mathematically presents sufficient guarantee global convergence NCELM. The update iteration defined contraction mapping function, through Banach theorem, proved.