Advanced Topics in Computer Science: Deep Generative Models:Methods, Applications & Societal Considerations

We study the core methods behind different classes of deep generative models. For each class of models, we study an important application it is useful for, for example generative adversarial networks and weather forecasting, or probabilistic diffusion models and molecule generation in 3D. We discuss the good uses and bad uses of these models, touching on their societal impact. There are guest lectures from academia and industry. Students are required to submit a brief report about the readings and conduct a semester-long research project.