Shaoan Xie | Carnegie Mellon University

Hi! I’m a PhD student at Carnegie Mellon University, advised by Prof. Kun Zhang and Prof. Peter Spirtes .
I have also been fortunate to intern at Adobe Research with Zhe Lin, Zhifei Zhang, and Tobias Hinz, and at Google with Yang Zhao, Zhisheng Xiao, and Kelvin C. K. Chan.

News

[Aug. 2025] I will be serving as an Area Chair for ICLR 2026 — thanks for the invitation!
[Jun. 2025] I’m attending CVPR 2025 — see you in Nashville!
[May. 2025] Two papers accepted into ICML 2025!
[Apr. 2025] One paper accepted as a CVPR 2025 Highlight!
[May. 2023] My work at Adobe, SmartBrush (CVPR 2023, Highlight), is now integrated into Adobe’s flagship products — Photoshop and Firefly!
[Jan. 2023] One paper, i-StyleGAN, accepted as an ICLR 2023 Spotlight! One of the first works to connect identifiability with multidomain image generation.

Research Highlights

My long-term research goal is to develop generative AI systems that go beyond producing realistic multimodal data to capturing the underlying structure of the world. By grounding machine learning and computer vision in principled representations of structure and dependencies, I aim to make generative models not only powerful, but also controllable, interpretable, trustworthy, and capable of systematic generalization.

Research Diagram

Causal Generative AI

How can we formulate guiding principles—particularly causality—to direct the learning of generative models?

How can we implement these principles in scalable generative systems?

Bridging theory and practice to build controllable generative AI

Theoretical Foundations

When and how true generative factors are identifiable from data.

Causal Representation Learning:

Counterfactual Reasoning:

Nonlinear Quantile Regression (arXiv 2024)

Controllable Visual Generation

Achieving minimal distortion while faithfully following conditions.

Multi-Domain Image Generation:

Identifiability Guarantees (ICLR 2023 Spotlight)

Unpaired Image Translation:

Shortest Path Regularization (CVPR 2023) Density Changing Regularization (NeurIPS 2022) Maximum Spatial Perturbation (CVPR 2022) Learning to Reweight (ICCV 2021)

Efficient Generation:

Semi-Implicit Denoising (SIDDMs) (NeurIPS 2023)

Vision-Language Alignment

Uncovering modular concepts from multimodal data.

Modular Concept Alignment:

SmartCLIP (CVPR 2025 Highlight)

Atomic Visual and Textual Concept for Control:

Vision & Language Concepts (ICML 2025)

Text-Guided Image Manipulation:

SmartBrush (CVPR 2023 Highlight) DreamInpainter (arXiv 2024)

Understanding and Post-Training

Opening the black box to interpret and refine generative models.

Interpretability of Generative Models:

Understanding Diffusion Models and LLMs through Hierachical Modeling

Policy Optimization for Generative Models:

Diffusion Language Models with Improved Reasoning Visual Generation Enhancement

Publications

arXiv

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

Shaoan Xie*, Lingjing Kong*, Xiangchen Song, Xinshuai Dong, Guangyi Chen, Eric P.Xing, Kun Zhang

(arXiv), 2025.

PDF
NeurIPS

Beyond the Black Box: Identifiable Interpretation and Control in Generative Models via Causal Minimality

Lingjing Kong*, Shaoan Xie*,, Guangyi Chen, Yuewen Sun, Xiangchen Song, Eric P. Xing, Kun Zhang

NeurIPS Mechanistic Interpretability Workshop, 2025.
arXiv

Controllable Video Generation with Provable Disentanglement

Yifan Shen*, Peiyuan Zhu*, Zijian Li, Shaoan Xie, Zeyu Tang, Namrata Deka, Zongfang Liu, Guangyi Chen, Kun Zhang

(arXiv), 2025.

PDF
ICML

Learning Vision and Language Concepts for Controllable Image Generation

Shaoan Xie*, Lingjing Kong*, Yujia Zheng, Zeyu Tang, Eric P.Xing, Guangyi Chen, Kun Zhang

International Conference on Machine Learning (ICML), 2025.

PDF Media
ICML

Nonparametric Identification of Latent Concepts

Yujia Zheng*, Shaoan Xie*, Kun Zhang

International Conference on Machine Learning (ICML), 2025.

PDF
CVPR

SmartCLIP: Modular Vision-language Alignment with Identification Guarantees

Shaoan Xie*, Lingjing Kong*, Yujia Zheng, Yu Yao, Zeyu Tang, Eric P.Xing, Guangyi Chen, Kun Zhang

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025. Highlight

PDF Code
AISTATS

Causal Representation Learning from General Environments under Nonparametric Mixing

Ignavier Ng, Shaoan Xie, Xingshuai Dong, Peter Spirtes, Kun Zhang

International Conference on Artificial Intelligence and Statistics (AISTATS), 2025.

PDF
ICLR

Synergy Between Sufficient Changes and Sparse Mixing Procedure for Disentangled Representation Learning

Zijian Li, Shunxing Fan, Yujia Zheng, Ignavier Ng, Shaoan Xie, Guangyi Chen, Xinshuai Dong, Ruichu Cai, Kun Zhang

International Conference on Learning Representations (ICLR), 2025.

PDF
ICML

Causal Representation Learning from Multiple Distributions: A General Setting

Kun Zhang*, Shaoan Xie*, Ignavier Ng*, Yujia Zheng

International Conference on Machine Learning (ICML), 2024.

PDF
arXiv

DreamInpainter: Text-Guided Subject-Driven Image Inpainting with Diffusion Models

Shaoan Xie, Yang Zhao, Zhisheng Xiao, Kelvin C.K. Chan, Yandong Li, Yanwu Xu, Kun Zhang, Tingbo Hou

(arXiv), 2024.

PDF
ICML

Advancing Counterfactual Inference through Nonlinear Quantile Regression

Shaoan Xie*, Biwei Huang*, Bin Gu, Tongliang Liu, Kun Zhang

ICML Workshop on Counterfactuals in Minds and Machines, 2024.

PDF
CVPR

Maximum Spatial Perturbation Consistency for Unpaired Image-to-Image Translation

Yanwu Xu, Shaoan Xie, Wenhao Wu, and Kun Zhang, Mingming Gong, Kayhan Batmanghelich

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022.

PDF Code
MICCAI

Adversarial Consistency for Single Domain Generalization in Medical Image Segmentation

Yanwu Xu, Shaoan Xie, Maxwell Reynolds, and Matthew Ragoza, Mingming Gong, Kayhan Batmanghelich

International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), 2022.

PDF
NeurIPS

Semi-Implicit Denoising Diffusion Models (SIDDMs)

Yanwu Xu, Mingming Gong, Shaoan Xie, Wei Wei, Matthias Grundmann, Kayhan Batmanghelich, Tingbo Hou

Advances in Neural Information Processing Systems (NeurIPS), 2023.

PDF
CVPR

SmartBrush: Text and Shape Guided Object Inpainting with Diffusion Model

Shaoan Xie, Zhifei Zhang, Zhe Lin, Tobias Hinz and Kun Zhang

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023. Highlight

PDF Video Alive in Adobe Photoshop
CVPR

Unpaired Image-to-Image Translation with Shortest Path Regularization

Shaoan Xie, Yanwu Xu, Mingming Gong and Kun Zhang

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023.

PDF Code Video
ICLR

Multi-domain Image Generation and Translation with Identifiability Guarantees

Shaoan Xie, Lingjing Kong, Mingming Gong and Kun Zhang

International Conference on Learning Representations (ICLR), 2023. Spotlight

PDF Code
NeurIPS

Unsupervised Image-to-Image Translation with Density Changing Regularization

Shaoan Xie, Qirong Ho and Kun Zhang

Advances in Neural Information Processing Systems (NeurIPS), 2022.

PDF Code
ICCV

Unaligned Image-to-Image Translation by Learning to Reweight

Shaoan Xie, Mingming Gong, Yanwu Xu and Kun Zhang

IEEE/CVF International Conference on Computer Vision (ICCV), 2021.

PDF Code
ICML

Learning Semantic Representations for Unsupervised Domain Adaptation

Shaoan Xie, Zibin Zheng, Liang Chen, and Chuan Chen

International Conference on Machine Learning (ICML), 2018.

PDF Code
ICML

Partial Disentanglement for Domain Adaptation

Lingjing Kong, Shaoan Xie, Weiran Yao, and Yujia Zheng, Guangyi Chen, Petar Stojanov, Victor Akinwande, Kun Zhang

International Conference on Machine Learning (ICML), 2022. Spotlight

PDF

Services

Area Chair

ICLR ’26

Session Chair

ICDM ’24

Conference Reviewer

ICML, NeurIPS, ICLR, CVPR, ICCV, UAI, AISTATS, IJCAI

Journal Reviewer

TPAMI, TIP, AI, CSUR, TNNLS, PR, TMM