Robust Data2VEC: Noise-Robust Speech Representation Learning for ASR by Combining Regression and Improved Contrastive Learning | IEEE Conference Publication | IEEE Xplore