Neural networks have become a popular choice in machine learning for modeling high dimensional, complex data generating processes. However, neural networks are black box models that come with significant barriers to explainability. One answer to this dilemma is to use Bayesian approaches to avoid potential issues with overfitting and measure model uncertainties. We will explore the theory behind Bayesian neural networks as well as the latest methods for fitting them.
- Introduction (5 min)
- Talk (~1 hr)
- Questions (5 min)
- Mingling (10 min)
Location: U Group
Address: 2231 Crystal Dr #401 · Arlington, VA
Parking: Free in the evening after 4pm at 2231 Crystal Drive the garage is below the building. Green Section.