Optimizing the Noise in Self-Supervised Learning: from Importance Sampling to Noise-Contrastive Estimation

Open in new window