Completed
Selective units
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Towards Falsifiable Interpretability Research in Machine Learning - Lecture
Automatically move to the next video in the Classroom when playback concludes
- 1 Introduction
- 2 Outline
- 3 Obstacles
- 4 Misdirection of Saliency
- 5 What is Saliency
- 6 Saliency axioms
- 7 Input invariants
- 8 Model parameter randomization
- 9 Does silencing help humans
- 10 Takeaways
- 11 Case Study 2
- 12 Individual neurons
- 13 Activation maximization
- 14 Populations
- 15 Selective units
- 16 Ablating selective units
- 17 Posthoc studies
- 18 Regularizing selectivity
- 19 Ingenerative models
- 20 Summary
- 21 Building better hypothesis hypotheses
- 22 Building a stronger hypothesis
- 23 Key takeaways
