Delves into score-based generative models, exploring learning natural distributions and the impact of neural network architecture on robustness.
Examines the generalization of ImageNet classifiers, safety-critical applications, overfitting, and the reliability of machine learning models.
Explores implicit generative models, covering topics like method of moments, kernel choice, and robustness of estimators.
Explores the challenges of robust vision, including distribution shifts, failure examples, and strategies for improving model robustness through diverse data pretraining.
