Exploring Group Video Captioning with Efficient Relational Approximation | IEEE Conference Publication | IEEE Xplore