Self-Supervised Learning¶
Self-supervised learning is the art of getting a model to teach itself from raw data before we ask it to solve a specific task. Instead of paying humans to label every image, we design a training signal from the data itself and use that to build a strong encoder.
Lectures¶
- Self-Supervised Learning: General principles of self-supervised learning.
- Contrastive Learning: Contrastive learning and CLIP.
- Self-Distillation: DINOv2