Lingjing Kong

Selection, Reflection and Self-Refinement: Revisit Reasoning Tasks via a Causal Lens

Identifiable Interpretation in Generative Models via Causal Minimality

Step-Aware Policy Optimization for Reasoning in Diffusion Large Language Models

Attribute-Guided Image Generation with Causally Disentangled Representation

Understanding Compositional Generalization via Hierarchical Concept Models

Towards Self-Refinement of Vision-Language Models with Triangular Consistency

Causal Representation Learning from Multimodal Biomedical Observations

Adjusting Pretrained Backbones for Performativity

Learning Vision and Language Concepts for Controllable Image Generation