I study interpreting modern generative models, with applications to 1) addressing misuse, alignment, and ethical problems in generative models, and 2) scaling generative models to more complex data, e.g. protein structures. I am more than fortunate to work with Prof. Jiaxuan You and Prof. Ge Liu.
Beyond the academia, I am deeply invested in helping disadvantaged groups, such as human artists, in the era of Generative AI. I co-found and lead a volunteering interest group that works on copyright issues of generative models. We develop softwares and provide technical services for AI copyright lawsuits as volunteers. E-mail us if you are interested!
Mist-v2 has released! Compared to Mist, Mist-v2 enjoys far more powerful protection performance against LoRA. We update its detailed features in the Homepage. We are also going to announce it in X. Looking forward to it!
May 10, 2023
Our project on adding adversarial watermarks against unauthorized artwork copying with Stable Diffusion, Mist, is now open-sourced on GitHub. Try to protect your artworks by adding tiny watermarks on them!
@article{wu2024cgi,title={CGI-DM: Digital Copyright Authentication for Diffusion Models via Contrasting Gradient Inversion},author={Wu, Xiaoyu and Hua, Yang and Liang, Chumeng and Zhang, Jiaru and Wang, Hao and Song, Tao and Guan, Haibing},journal={arXiv preprint arXiv:2403.11162},year={2024}}
ICLR
Toward effective protection against diffusion based mimicry through score distillation
Haotian Xue, Chumeng Liang, Xiaoyu Wu, and 1 more author
@inproceedings{xue2023toward,title={Toward effective protection against diffusion based mimicry through score distillation},author={Xue, Haotian and Liang, Chumeng and Wu, Xiaoyu and Chen, Yongxin},booktitle={arXiv preprint arXiv:2311.12832},year={2024}}
2023
ICMLOral
Adversarial Example Does Good: Preventing Painting Imitation from Diffusion Models via Adversarial Examples
Chumeng Liang*, Xiaoyu Wu*, Yang Hua, and 6 more authors
In Proceedings of the 40th International Conference on Machine Learning, 2023
@inproceedings{pmlr-v202-liang23g,tag={Oral},title={Adversarial Example Does Good: Preventing Painting Imitation from Diffusion Models via Adversarial Examples},author={Liang*, Chumeng and Wu*, Xiaoyu and Hua, Yang and Zhang, Jiaru and Xue, Yiming and Song, Tao and Xue, Zhengui and Ma, Ruhui and Guan, Haibing},booktitle={Proceedings of the 40th International Conference on Machine Learning},pages={20763--20786},year={2023}}
KDD
CBLab: Supporting the Training of Large-scale Traffic Control Policies with Scalable Traffic Simulation
Chumeng Liang, Zherui Huang, Yicheng Liu, and 7 more authors
In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2023
@inproceedings{liang2023cblab,title={CBLab: Supporting the Training of Large-scale Traffic Control Policies with Scalable Traffic Simulation},author={Liang, Chumeng and Huang, Zherui and Liu, Yicheng and Liu, Zhanyu and Zheng, Guanjie and Shi, Hanyuan and Wu, Kan and Du, Yuhao and Li, Fuliang and Li, Zhenhui Jessie},booktitle={Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining},pages={4449--4460},year={2023}}
