Machine learning approaches for classification of myriad underwater acoustic events over continental-shelf scale regions with passive ocean acoustic waveguide remote sensing