The remarkable success stories of deep learning systems give the impression that large advanced neural networks are all we need, and they will eventually make all the symbolic techniques obsolete. This predicted death of symbolic techniques might or might not happen in the end. But I believe that at least in the next 5-10 years, symbolic techniques will continue to have something to offer to neural networks especially in the reasoning tasks. In this survey talk, I will review three proposals for using symbolic techniques to build deep-learning systems that can reason.
