Zhengyuan Su (苏政渊)

National University of Singapore

Hello, I am Zhengyuan Su, a first-year PhD student in Computer Science at National University of Singapore (NUS), advised by Prof. Yao Lu (https://yao.lu/). I got my Bachelor degree from the Yao Class at Tsinghua University.

My research interest lies broadly in artificial intelligence, especially the intersection of Machine Learning and systems. I am also interested in cryptography. I have worked on the following projects:

The ColTrain project, working on outlier-aware LLM quantization.
Syno, an ML operator synthesizer that synthesizes efficient neural network kernels with minimal accuracy degradation.
FABLE, a cryptographic protocol for confidential LUT evaluation.
An ML project about multimodal learning between 3D computer vision and natural language processing.

I am fortunate to collaborate with Prof. Mingyu Gao and Prof. Li Yi at Tsinghua University, Prof. Wenting Zheng at CMU, and Prof. Babak Falsafi at EPFL.

news

Aug 30, 2025	Our USENIX Security paper “FABLE: Batched Evaluation on Confidential Lookup Tables in 2PC” (me as the first author) is published on USENIX Security ‘25!
Aug 6, 2025	Our ASPLOS paper “Syno: Structured Synthesis for Neural Operators” (me as a co-first author) is published on ASPLOS ‘25!
Aug 23, 2024	I was awarded Yao Award by IIIS!
Jun 29, 2023	Our paper “Language-Assisted 3D Feature Learning for Semantic Scene Understanding” is available on AAAI’s official archive.
Nov 19, 2022	Our paper “Language-Assisted 3D Feature Learning for Semantic Scene Understanding” (also my first publication) is accepted to AAAI 2023.

selected publications

2025

ASPLOS ’25

Syno: Structured Synthesis for Neural Operators

Yongqi Zhuo, Zhengyuan Su, Chenggang Zhao, and 1 more author

In Proceedings of the 30th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 3, 2025

arXiv Bib Code

@inproceedings{zhuo2025syno,
  author = {Zhuo, Yongqi and Su, Zhengyuan and Zhao, Chenggang and Gao, Mingyu},
  title = {Syno: Structured Synthesis for Neural Operators},
  year = {2025},
  isbn = {9798400710803},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  url = {https://doi.org/10.1145/3676642.3736118},
  doi = {10.1145/3676642.3736118},
  booktitle = {Proceedings of the 30th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 3},
  pages = {212–229},
  numpages = {18},
  keywords = {neural architecture search, program synthesis},
  location = {Rotterdam, Netherlands},
  series = {ASPLOS '25},
}

USENIX Sec ’25

FABLE: Batched Evaluation on Confidential Lookup Tables in 2PC

Zhengyuan Su, Qi Pang, Simon Beyzerov, and 1 more author

2025

Bib Code

@misc{su2025fable,
  author = {Su, Zhengyuan and Pang, Qi and Beyzerov, Simon and Zheng, Wenting},
  title = {{FABLE}: Batched Evaluation on Confidential Lookup Tables in {2PC}},
  howpublished = {Cryptology {ePrint} Archive, Paper 2025/1081},
  year = {2025},
  eprint = {https://eprint.iacr.org/2025/1081},
}

2023

AAAI ’23

Language-Assisted 3D Feature Learning for Semantic Scene Understanding

Junbo Zhang, Guofan Fan, Guanghan Wang, and 3 more authors

In Proceedings of the AAAI Conference on Artificial Intelligence, 2023

Bib HTML

@inproceedings{zhang2023language,
  title = {Language-Assisted 3D Feature Learning for Semantic Scene Understanding},
  author = {Zhang, Junbo and Fan, Guofan and Wang, Guanghan and Su, Zhengyuan and Ma, Kaisheng and Yi, Li},
  booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  volume = {37},
  number = {3},
  pages = {3445--3453},
  year = {2023},
  doi = {10.1609/aaai.v37i3.25453},
}