On the Connection Between Neural Networks and Kernels: A Modern Perspective - Simon Du

On the Connection Between Neural Networks and Kernels: A Modern Perspective - Simon Du

Institute for Advanced Study via YouTube Direct link

Intro

1 of 24

1 of 24

Intro

Class Central Classrooms beta

YouTube playlists curated by Class Central.

Classroom Contents

On the Connection Between Neural Networks and Kernels: A Modern Perspective - Simon Du

Automatically move to the next video in the Classroom when playback concludes

  1. 1 Intro
  2. 2 Two Fundamental Questions
  3. 3 Empirical Observations on Training Loss
  4. 4 Over-parameterization
  5. 5 Empirical Observations on Generalization
  6. 6 Example: Two-layer NN
  7. 7 Trajectory-based Analysis
  8. 8 The Trajectory of Predictions (Cont'd)
  9. 9 Kernel Matrix at the Beginning
  10. 10 Kernel Matrix During Training
  11. 11 Main Theory
  12. 12 Zero Training Error
  13. 13 Empirical Results on Generalization
  14. 14 Convolutional Neural Tangent Kernel
  15. 15 CNTK on CIFAR 10
  16. 16 Understanding Global Average Pooling
  17. 17 Local Average Pooling
  18. 18 UCI Experiment Setup
  19. 19 UCI Results
  20. 20 Few-shot Learning Setup
  21. 21 Few-shot Learning Results
  22. 22 Graph NTK for Graph Classification
  23. 23 Summary
  24. 24 References

Never Stop Learning.

Get personalized course recommendations, track subjects and courses with reminders, and more.

Someone learning on their laptop while sitting on the floor.