teaching
Back in grad school, I was a teaching assistant for Carnegie Mellon’s convex optimization course. We covered:
- first-order methods
- second-order methods
- interior point methods
- momentum and acceleration
- stochastic optimization
- distributed optimization
- non-convex optimization (e.g., quasi-convex optimization, mixed integer programming, …)
- contact points with statistics (e.g., implicit regularization)
- some convex analysis
Somewhat more recently, I taught an (extremely talented) group of high schoolers as part of Stanford’s “STEM to SHTEM” summer program. We covered:
- the basics of supervised learning
- how to build a (modern) recommendation system
- matrix completion / factorization-based methods
- neighborhood-based methods
- prediction-based methods
- bandits
- uncertainty quantification
- evaluation