Are All Features Created Equal? - Aleksander Madry

Are All Features Created Equal? - Aleksander Madry

Institute for Advanced Study via YouTube Direct link

Intro

1 of 22

1 of 22

Intro

Class Central Classrooms beta

YouTube playlists curated by Class Central.

Classroom Contents

Are All Features Created Equal? - Aleksander Madry

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

  1. 1 Intro
  2. 2 Machine Learning: A Success Story
  3. 3 Why Do We Love Deep Learning?
  4. 4 Key Phenomenon: Adversarial Perturbations
  5. 5 ML via Adversarial Robustness Lens
  6. 6 But: "How"/"what" does not tell us "why"
  7. 7 Why Are Adv. Perturbations Bad?
  8. 8 Human Perspective
  9. 9 ML Perspective
  10. 10 A Simple Experiment
  11. 11 The Robust Features Model
  12. 12 The Simple Experiment: A Second Look
  13. 13 Human vs ML Model Priors
  14. 14 New capability: Robustification
  15. 15 Some Direct Consequences
  16. 16 Robustness and Data Efficiency
  17. 17 Robustness + Perception Alignment
  18. 18 Robustness → Better Representations
  19. 19 Robustness + Image Synthesis
  20. 20 Problem: Correlations can be weird
  21. 21 Useful tool(?): Counterfactual Analysis with Robust Models
  22. 22 Adversarial examples arise from non-robust features in the data

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.