Are Gaussian Data All You Need for Machine Learning Theory - A Statistical Physics Perspective

Are Gaussian Data All You Need for Machine Learning Theory - A Statistical Physics Perspective

Harvard CMSA via YouTube Direct link

Intro

1 of 23

1 of 23

Intro

Class Central Classrooms beta

YouTube videos curated by Class Central.

Classroom Contents

Are Gaussian Data All You Need for Machine Learning Theory - A Statistical Physics Perspective

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

  1. 1 Intro
  2. 2 The machine learning revolutic
  3. 3 Gaussians data
  4. 4 Physicists & Theoretical Neuroscie
  5. 5 Random Matrix Theory, Statistics
  6. 6 Data agnostic approaches
  7. 7 Fitting real dataset with a Gaussian model
  8. 8 Ridge regression on MNIST
  9. 9 But Gaussian theory does not always work
  10. 10 Ridge, Logistic, Hinge classification vs Gaussian
  11. 11 What is a better model than a single Gaussian?
  12. 12 Theorem: Gaussian Mixture through random
  13. 13 Open problems: beyond proportional regim
  14. 14 Generative Neural Networks as a proxy for r
  15. 15 GAN generated data behaves as Gaussian Mi
  16. 16 GMM stays GMM through random features
  17. 17 Theorem: Asymptotic of the Gaussian Mixture mo
  18. 18 2 Gaussians vs 1 Gaussian for different teacher
  19. 19 Single Gaussian for randon
  20. 20 Remember this plot with random label?
  21. 21 Universality of phase transition for homoskedast
  22. 22 Ridge interpolator & random
  23. 23 Many more questions...

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.