Yao Xiao | Sun Yat-sen University

About Me

I am an EngD student at Sun Yat-sen University, supervised by Prof. Liang Lin. I am a member of HCP Lab (Human-Cyber-Physical Intelligence Integration Lab) leaded by Prof. Liang Lin, and currently a visiting student in LV-Lab, National University of Singapore, under the supervision of Prof. Shuicheng Yan. Before that, I received the MEng and BEng in Computer Science and Technology at Sun Yat-sen University in 2023 and 2020, respectively, advised by Prof. Liang Lin, Prof. Cong Liu, and Prof. Pengxu Wei.

Research Interests

My general research interest is to build trustworthy and evolving deep learning systems for computer vision and beyond, including:

AI Generated Content (AIGC): detection/evaluation of AIGC, controllable and efficient image/video generation
Machine Learning: adversarial robustness, domain adaptation, causality in deep learning

News

[May 2025] One paper on AI-generated image detection accepted to ICML 2025.
[Jan. 2025] One paper on adversarial robustness accepted to ICLR 2025.
[July 2024] One paper on controllable image generation accepted to ACMMM 2024.
[Mar. 2023] One paper on adversarially robust source-free domain adaptation is accepted to ICME 2023 (oral).
[Mar. 2023] One paper on large model fine-tuning and causality is accepted to CVPR 2023.

Publications

ICML

Are High-Quality AI-Generated Images More Difficult for Models to Detect?

Yao Xiao, Binbin Yang, Weiyan Chen, Jiahao Chen, Zijie Cao, Ziyi Dong, Xiangyang Ji, Liang Lin, Wei Ke, Pengxu Wei

Forty-second International Conference on Machine Learning (ICML), 2025.

PDF Code

ICLR

Towards Understanding the Robustness of Diffusion-Based Purification: A Stochastic Perspective

Yiming Liu*, Kezhao Liu*, Yao Xiao, Ziyi Dong, Xiaogang Xu, Pengxu Wei, Liang Lin

The Thirteenth International Conference on Learning Representations (ICLR), 2025.

PDF Code

ACMMM

Decoder-Only LLMs are Better Controllers for Diffusion Models

Ziyi Dong, Yao Xiao, Pengxu Wei, Liang Lin

Proceedings of the 32nd ACM International Conference on Multimedia (ACMMM), 2024.

PDF

ICME

Adversarially Robust Source-free Domain Adaptation with Relaxed Adversarial Training

Yao Xiao, Pengxu Wei, Cong Liu, Liang Lin

IEEE International Conference on Multimedia and Expo (ICME), 2023.

PDF Code Oral

CVPR

Masked Images Are Counterfactual Samples for Robust Fine-tuning

Yao Xiao, Ziyi Tang, Pengxu Wei, Cong Liu, Liang Lin

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

PDF Code

Services

I serve as conference reviewer for ICML, ICLR, NeurIPS, AAAI, ACMMM, and journal reviewer for IEEE TNNLS, IEEE TIP.