Week 9

Lecture

• Lecture #16: Neural Network Models for Regression
  • Lecture Video
  • Lecture Notes
  • Lecture Summary
• Lecture #17: Black-box Variational Inference
  • Lecture Video
  • Lecture Notes
  • Lecture Summary

Activities

• In-Class Exercise #16: Understanding Neural Networks
• In-Class Exercise #16: Solutions
• In-Class Exercise #17: Black-box Variational Inference
• In-Class Exercise #17: Solutions
• Homework #8

Reading

Applications and Broader Impact:

1. The Mythos of Model Interpretability
2. Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead

Variational Inference for Bayesian Neural Networks

1. Black Box Variational Inference
2. Weight Uncertainty in Neural Networks
3. Sticking the Landing: Simple, Lower-Variance Gradient Estimators for Variational Inference