A neuro-dynamic architecture for one shot learning of objects that uses both bottom-up recognition and top-down prediction | IEEE Conference Publication | IEEE Xplore