Boosting ridge extreme learning machine

Extreme Learning Machine (ELM) was proposed recently for efficiently training single-hidden-layer feed forward neural networks (SLFNs). However, this algorithm often requires a large number of hidden nodes and sometimes performs unstable. In this paper, a novel training algorithm called boosting ridge extreme learning machine (BR-ELM) is proposed which can construct stable generalization performance with a compact network with boosting ridge regression. In the experiments, the BR-ELM remains robust to all the test data and performs stable and response fast with much less nodes.