Efficient Speaker Naming via Deep Audio-Face Fusion and End-to-End Attention Model | IEEE Conference Publication | IEEE Xplore