Explores stationarity in stochastic processes, showcasing how statistical characteristics remain constant over time and the implications on random variables and Fourier transforms.
Explores the challenges of robust vision, including distribution shifts, failure examples, and strategies for improving model robustness through diverse data pretraining.