Gege Gao

Statistical dependences underlie machine learning, and their detection in large-scale i.i.d. settings has led to impressive results in a range of domains, but in many situations, we would prefer a causal model to a purely predictive one. Current methods of causal inference have certain weakness that this project will try to address: it ignores the problem of how to infer causal representations from raw data like images. In particular, the PhD student will work on novel learning-based models that provide insights into robust inference of causal representations from visual data.