Multi-Task Corrupted Prediction for Learning Robust Audio-Visual Speech Representation.

Sungnyun Kim, Sungwoo Cho, Sangmin Bae, Kangwook Jang, Se-Young Yun

Browse the full ICLR paper archive.