Self-Supervised Learning of Visual Features through Embedding Images into Text Topic Spaces | IEEE Conference Publication | IEEE Xplore