Chaoran Cheng

Hi! I am Chaoran Cheng, a third-year CS Ph.D. candidate at the University of Illinois Urbana-Champaign. I am advised by Prof. Ge Liu, before which I was working with Prof. Jian Peng. My research interests mainly span multiple aspects of generative modeling in AI4Science domains, with a special interest in manifold learning in which desirable network behaviors are mathematically inspired and guaranteed. Combining continuous and discrete modalities with a more comprehensive and versatile generative model for molecules and proteins, our work can build a more comprehensive and versatile generative model and has been successfully applied to various AI4Science tasks including protein sequence-structure co-design and structure-based drug design.

I am actively engaged in the MMLI initiative, a joint effort between UIUC CS School and Chemistry Department to develop a machine learning-based platform for molecular design. I am also engaged in the CABBI initiative, a multi-institutional project to deal with the emergent challenges of sustainable bioproducts.

I obtained my Bachelor’s Degree in Computer Science at Peking University, before which I also spent my freshman year studying chemistry.

Other personal interests

Out of my pure personal interest, I am developing a machine-learning based automatic charter for Cytoid, a community-driven rhythm game inspired by Rayark's games Cytus and Cytus II. The community provides many fan-made charts that can be used to train a machine learning system.

News

Nov 6, 2024	My contributing work on AI²BMD during my internship at MSRA has been accepted to Nature. In this work, we extend AI-assisted molecular dynamics (MD) simulation to large-scale protein systems with ab initio accuracy. Check out MSR's introductory blog on why MD can be fundamental to various AI4Science domains.
Sep 27, 2024	Our latest work of Statistical Flow Matching (SFM) for discrete generation has been accepted to NeurIPS 2024! Check out our paper on arXiv and our code on GitHub. We applied SFM across image, text, and biological domains demonstrate its superior sampling quality and NLL on discrete tasks. (1/4) Can we unleash the power of Riemannian Flow Matching on discrete generation? We introduce “Statistical Flow Matching (SFM)”, a novel FM framework on the manifold of parameterized probability measures inspired by information geometry. Check it out at https://t.co/7aTdwkcNvk pic.twitter.com/QoVkn1mB8I— Chaoran Cheng (@ccr_cheng) May 28, 2024
Sep 21, 2023	My work Equivariant Neural Operator Learning with Graphon Convolution was accepted to NeurIPS 2023 as a Spotlight paper. Check out the paper here. I am also honored to be the Workflow Chair of the New Frontiers in Graph Learning Workshop at NeurIPS 2023!
Dec 7, 2022	Check out this post on the Tensor Field Network! ⌬
Nov 30, 2022	Get a glimpse of Cézanne and the revolution of modern art.

Selected Publications

Ab initio characterization of protein molecular dynamics with AI2BMD

Tong Wang, Xinheng He, Mingyu Li, Yatao Li, Ran Bi, Yusong Wang, Chaoran Cheng, Jiawei Shen, He Zhang, Haiguang Liu, Zun Wang, Shaoning Li, Bin Shao, and Tie-Yan Liu

Nature Nov 2024

Abs Bib HTML Code

Biomolecular dynamics simulation is a fundamental technology for life sciences research, and its usefulness depends on its accuracy and efficiency1–3. Classical molecular dynamics simulation is fast but lacks chemical accuracy4,5. Quantum chemistry methods such as density functional theory can reach chemical accuracy but cannot scale to support large biomolecules6. Here we introduce an artificial intelligence-based ab initio biomolecular dynamics system (AI2BMD) that can efficiently simulate full-atom large biomolecules with ab initio accuracy. AI2BMD uses a protein fragmentation scheme and a machine learning force field7 to achieve generalizable ab initio accuracy for energy and force calculations for various proteins comprising more than 10,000 atoms. Compared to density functional theory, it reduces the computational time by several orders of magnitude. With several hundred nanoseconds of dynamics simulations, AI2BMD demonstrated its ability to efficiently explore the conformational space of peptides and proteins, deriving accurate 3J couplings that match nuclear magnetic resonance experiments, and showing protein folding and unfolding processes. Furthermore, AI2BMD enables precise free-energy calculations for protein folding, and the estimated thermodynamic properties are well aligned with experiments. AI2BMD could potentially complement wet-lab experiments, detect the dynamic processes of bioactivities and enable biomedical research that is impossible to conduct at present.
@article{Wang2024, author = {Wang, Tong and He, Xinheng and Li, Mingyu and Li, Yatao and Bi, Ran and Wang, Yusong and Cheng, Chaoran and Shen, Xiangzhen an d Meng, Jiawei and Zhang, He and Liu, Haiguang and Wang, Zun and Li, Shaoning and Shao, Bin and Liu, Tie-Yan}, title = {Ab initio characterization of protein molecular dynamics with AI2BMD}, journal = {Nature}, year = {2024}, month = nov, day = {06}, issn = {1476-4687}, doi = {10.1038/s41586-024-08127-z}, url = {https://doi.org/10.1038/s41586-024-08127-z}, }

Categorical Flow Matching on Statistical Manifolds

Chaoran Cheng, Jiahan Li, Jian Peng, and Ge Liu

In Annual Conference on Neural Information Processing Systems 2024, NeurIPS 2024 Nov 2024

Bib HTML Code

@inproceedings{cheng2024categorical,
  title = {Categorical Flow Matching on Statistical Manifolds},
  author = {Cheng, Chaoran and Li, Jiahan and Peng, Jian and Liu, Ge},
  year = {2024},
  booktitle = {Annual Conference on Neural Information Processing Systems 2024, NeurIPS 2024},
}

Neural P³M: A Long-Range Interaction Modeling Enhancer for Geometric GNNs

Yusong Wang, Chaoran Cheng, Shaoning Li, Yuxuan Ren, Bin Shao, Ge Liu, Pheng-Ann Heng, and Nanning Zheng

In Annual Conference on Neural Information Processing Systems 2024, NeurIPS 2024 Nov 2024

Bib HTML

@inproceedings{wang2024neuralp3m,
  title = {Neural P³M: A Long-Range Interaction Modeling Enhancer for Geometric GNNs},
  author = {Wang, Yusong and Cheng, Chaoran and Li, Shaoning and Ren, Yuxuan and Shao, Bin and Liu, Ge and Heng, Pheng-Ann and Zheng, Nanning},
  year = {2024},
  booktitle = {Annual Conference on Neural Information Processing Systems 2024, NeurIPS 2024},
}

Full-Atom Peptide Design based on Multi-modal Flow Matching

Jiahan Li, Chaoran Cheng, Zuofan Wu, Ruihan Guo, Shitong Luo, Zhizhou Ren, Jian Peng, and Jianzhu Ma

In Forty-first International Conference on Machine Learning, ICML 2024 Nov 2024

Bib HTML

@inproceedings{li2024fullatom,
  title = {Full-Atom Peptide Design based on Multi-modal Flow Matching},
  booktitle = {Forty-first International Conference on Machine Learning, {ICML} 2024},
  author = {Li, Jiahan and Cheng, Chaoran and Wu, Zuofan and Guo, Ruihan and Luo, Shitong and Ren, Zhizhou and Peng, Jian and Ma, Jianzhu},
  series = {Proceedings of Machine Learning Research},
  publisher = {{PMLR}},
  year = {2024},
}

Equivariant Neural Operator Learning with Graphon Convolution

Chaoran Cheng, and Jian Peng

In Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023 Nov 2023

Bib PDF Code

@inproceedings{Cheng2023infgcn,
  title = {Equivariant Neural Operator Learning with Graphon Convolution},
  author = {Cheng, Chaoran and Peng, Jian},
  booktitle = {Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023},
  year = {2023},
}

Equivariant Point Cloud Analysis via Learning Orientations for Message Passing

Shitong Luo, Jiahan Li, Jiaqi Guan, Yufeng Su, Chaoran Cheng, Jian Peng, and Jianzhu Ma

In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Jun 2022

Bib PDF Code

@inproceedings{Luo2022CVPR,
  author = {Luo, Shitong and Li, Jiahan and Guan, Jiaqi and Su, Yufeng and Cheng, Chaoran and Peng, Jian and Ma, Jianzhu},
  title = {Equivariant Point Cloud Analysis via Learning Orientations for Message Passing},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = jun,
  year = {2022},
  pages = {18932-18941},
}