sp26-cs639 lectureself-supervised-learningcontrastive-learninggnncs639
Lecture 5: Self-Supervised Learning Lecture 5: Self-Supervised Learning
Content locked
The catalog is public. Reading this content requires approved access.
Sign in to readLecture 5: Self-Supervised Learning
Sequence Models & Graph Models
Tasks We Can Handle with NNs
Simple RNNs
LSTMs
Transformers
Relationships in Data
Graph Neural Networks: Motivations
Graph Neural Networks: Approach
Graph Convolutional Networks (GCNs)
Self-Supervised Learning
Representations
Where to Get Representations?
Representations from Self Supervision
Using the Representations
Pretext Tasks
Teacher-Student Methods
Contrastive Learning
Using Multiple Samples
Training with Positive and Negative Samples
Contrastive Losses
Triplet Loss
InfoNCE Loss
Generating Positives
Generating Negatives
Putting it Together: Contrastive Training
Frameworks