Lingjing Kong

This work highlights dependencies between content and style in counterfactual generation.

This study proposes a framework for understanding hidden concepts in complex data.