1 paper
Aditya Radhakrishnan, Maxwell A. Xu, Wanting Mao +3
Contrastive learning for time series depends on deciding which pairs of inputs should be treated as similar and dissimilar. Because few augmentations reliably preserve meaning in t…