TAQ: Top-K Attention-Aware Quantization for Vision Transformers | IEEE Conference Publication | IEEE Xplore