A Multiple Classifier-based Concept-Spotting Approach for Robust Spoken Language Understanding