Semantic Compositional Networks for Visual Captioning | IEEE Conference Publication | IEEE Xplore