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