To do interesting things with joint distributions, they need to have dependence structure. This is part of a series of videos for COS 302: Mathematics for Numerical Computation and Machine Learning, replacing lectures after the course went remote due to the COVID-19 pandemic.