One of the most promising applications of cross-modal machine learning is neural search. The core idea of neural search is to leverage state-of-the-art deep neural networks to build every component of a search system.