Implicit Generative Models

This lecture covers the theory behind implicit generative models, focusing on topics such as generative models, method of moments, integral probability metrics, reproducing kernel Hilbert space, maximum mean discrepancy, minimum MMD estimators, kernel choice in high-dimensional models, asymptotic normality, efficiency of estimators, robustness of minimum MMD estimators, and speeding up gradient descent. The instructor discusses the challenges and methodologies related to learning implicit generative models, providing insights into inference for ordinary and stochastic differential equations, as well as stochastic reaction networks.