Qing Qu

Assistant Professor@University of Michigan·美国·OpenReview

研究方向

Nonconvex optimization · Representation Learning · Deep Learning Theory · Low-dimensional Modeling · AI4Science · Deep Generative Models · Foundations of Generative AI

Qing Qu

AlphaFlow: Understanding and Improving MeanFlow Models

Generalization of Diffusion Models Arises with a Balanced Representation Space

Unlearning Isn't Invisible: Detecting Unlearning Traces in LLMs from Model Outputs

The Dual Power of Interpretable Token Embeddings: Jailbreaking Attacks and Defenses for Diffusion Model Unlearning

Towards Understanding the Mechanisms of Classifier-Free Guidance

A Closer Look at Model Collapse: From a Generalization-to-Memorization Perspective

Understanding Representation Dynamics of Diffusion Models via Low-Dimensional Modeling

FlowDAS: A Stochastic Interpolant-based Framework for Data Assimilation

Learning Dynamics of Deep Matrix Factorization Beyond the Edge of Stability

Shallow Diffuse: Robust and Invisible Watermarking through Low-Dim Subspaces in Diffusion Models

UGoDIT: Unsupervised Group Deep Image Prior Via Transferable Weights

Attention-Only Transformers via Unrolled Subspace Denoising

SITCOM: Step-wise Triple-Consistent Diffusion Sampling For Inverse Problems

Shallow Diffuse: Robust and Invisible Watermarking through Low-Dimensional Subspaces in Diffusion Models

Attention-Only Transformers via Unrolled Subspace Denoising

Model Collapse Analysis and Improvement for Rectified Flow Models

Step-wise Triple-Consistent Diffusion Sampling for Inverse Problems

Understanding Diffusion-based Representation Learning via Low-Dimensional Modeling

Diffusion Models Learn Low-Dimensional Distributions via Subspace Clustering