Completed
TPR disparities persist in "balanced" datasets
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Adversarial Debiasing With Partial Learning - Medical Image Studies
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Bias in ML models
- 3 Background: Studying Biases in Medical AI models
- 4 These Algorithms Look at X-Rays-and Somehow Detect Your Race
- 5 Solution: Retrain Model With Balance Dataset
- 6 Emory CXR dataset Balanced Training
- 7 TPR disparities persist in "balanced" datasets
- 8 Adversarial debiasing: Unlearn Biasing features
- 9 Adversarial Debiasing Architecture
- 10 Adversarial Debiasing Background
- 11 Ablation Studies: Review
- 12 How do we identify the layers to debias?
- 13 Emory Mammogram Dataset • Cohorts of 150-180k patients each featuring screening
- 14 Emory Mammography Dataset (Race Distribution )
- 15 Deep learning model for tissue density classification
- 16 Studying Models Ability to predict Race
- 17 Findings on Predicting Race
- 18 Removing Race Related Features (Step 2)
- 19 TPR Disparity Measures
- 20 CXR Model Performance
- 21 Questions?
- 22 CXR Debiasing Results
