Yixin Wang, Assistant Professor, Statistics, University of Michigan
Representation learning constructs low-dimensional representations to summarize essential features of high-dimensional data like images and texts. Ideally, such a representation should efficiently capture non-spurious features of the data. It shall also be disentangled so that we can interpret what feature each of its dimensions capture. However, these desiderata are often intuitively defined and challenging to quantify or enforce.
In this talk, we take on a causal perspective of representation learning. We show how desiderata of representation learning can be formalized using counterfactual notions, enabling metrics and algorithms that target efficient, non-spurious, and disentangled representations of data. We discuss the theoretical underpinnings of the algorithm and illustrate its empirical performance in both supervised and unsupervised representation learning.