PhD student, Department of Computer Science
1 paper at NeurIPS 2025
AdaTS introduces a dynamic soft contrastive learning framework for adaptive time series representations, leveraging frequency-domain similarities with dynamic temporal and ordinal instance contrasts.
