ManySpace
Türkçe Sign in Sign up

AI & MLCoreAIML-EMB-008

Contrastive Learning

Pull matched pairs together, push others apart: InfoNCE as softmax cross-entropy over similarities.

Difficulty
2 / 5
Importance
5 / 5
Study time
1–2 h
Kind of work
Mixed

Why it matters

The learner can compute an InfoNCE loss for a small batch and explain how temperature and negatives shape the embedding space.

The route to this topic

Route AIML-EMB-008
What Is a Model? Tasks: Regression, Classification, and Beyond Features and Targets Learning Paradigms: Where the Supervision Signal Comes From Classification Framing Linear Regression I: Prediction and Residuals Embeddings Scores and Logits Self-Supervised Learning Loss and Fitting: Squared Error Similarity in Vector Space Multiclass Scores and Softmax Objective Functions and Empirical Risk Likelihood Cross-Entropy Loss Contrastive Learning