Completed
Assumptions
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Towards Better Understanding of Contrastive Learning
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Great Empirical Success of Deep Models
- 3 Self-supervised Learning (SSL)
- 4 Contrastive Learning (CL)
- 5 Formulation of Contrastive Learning
- 6 Understanding of Contrastive Loss
- 7 What Deep Learning Brings?
- 8 Example: InfoNCE
- 9 Coordinate-wise Optimization
- 10 A Surprising Connection to Kernels
- 11 Overview of Nonlinear Analysis
- 12 Nonlinear Setting
- 13 Training Dynamics
- 14 1-layer 1-node nonlinear network
- 15 How to reduce the local roughness p(w)?
- 16 1-layer multiple node nonlinear network
- 17 Assumptions
- 18 Conditional Independence
- 19 What linear network cannot do
- 20 Global modulation
- 21 Feature Emergence
- 22 Experiment Setting
- 23 Model Architecture & Evaluation Metric
- 24 Visualization
- 25 Quadratic Loss versus InfoNCE
