We propose using support vector machines (SVMs) to learn the efficient set in multiple objective discrete optimization (MODO). We conjecture that a surface generated by SVM could provide a good approximation of the efficient set. As one way of testing this idea, we embed the SVM-approximated...