Zhe Zeng 曾哲

Assistant Professor

Hi there, I'm Zhe Zeng. I am an Assistant Professor in the Department of Computer Science at the University of Virginia, directing the Trustworthy AI through Knowledge and Optimization (TAKO

) Lab. My research centers on neurosymbolic AI and probabilistic machine learning to achieve reliable, interpretable and trustworthy AI. Prior to joining UVA, I was a Faculty Fellow in the Computer Science Department at New York University (NYU) hosted by Prof. Andrew Gordon Wilson. I obtained my Ph.D. degree in Computer Science at University of California, Los Angeles (UCLA), where I was lucky to be advised by Prof. Guy Van den Broeck.

📢 Joining

We're hiring! If you're interested in working with us, please fill out this interest form. I will read all submissions, though I may not be able to reply to each individually.

Prospective PhD Students: I am actively recruiting PhD students. Please apply to the UVA Computer Science PhD program and mention my name in your application as a preferred advisor.

Postdocs: Please fill in the form and I will contact you for a quick chat.

Prospective Research Interns: Take a look at our group webpage to see which research directions interest you. If you're a UVA undergrad or master's student, feel free also to reach out to our PhD students directly to discuss potential projects.

Visitors: Feel free to reach out if our interests align.

⚡ Research

My research interests lie broadly in artificial intelligence (AI) and machine learning (ML) with recent focus on neurosymbolic AI and probabilistic ML. We aim to enable and support decision-making in the real world in the presence of probabilistic uncertainty and symbolic knowledge (graph structures, logical, arithmetic, and physical constraints, etc) to achieve trustworthy AI and aid scientific discoveries. Our current work can be roughly catalogued as:

Reasoning: tractable probabilistic inference, weighted model integration, probabilistic circuits, and hybrid inference with logical & arithmetic constraints
Learning: constrained deep generative models, gradient estimators for discrete/combinatorial constraints, graph machine learning
Trustworthiness: uncertainty quantification and Bayesian deep learning, reliable LLMs (hallucination detection, constrained decoding)

🌟 News

Aug 2026 funding

We received the DARPA CLARA grant! ✨
Apr 2026 paper

Two papers accepted to ICML 2026: ProbMoE: Differentiable Probabilistic Routing for Mixture-of-Experts and Automatic Layer Selection for Hallucination Detection 🎉
Apr 2026 paper

One paper accepted to NeSy 2026: Learning Deep Generative Models under Hard Linear Equality Constraints 🎉
Feb 2026 funding

We received the UVA DAC Analytics Resource Award! ✨
Aug 2025 team

Welcome Zhizhen, Heng and Xinpeng to the group 🥳

💡 Recent Publications

All publications →

Entropy-Guided LLM Decoding via Probabilistic Circuits

Zhizhen Chen, Daniel Mingyi Israel, Guy Van den Broeck, Zhe Zeng

UAI 2026 Workshop TPM , 2026

PDF

@inproceedings{
chen2026entropyguided,
title={Entropy-Guided {LLM} Decoding via Probabilistic Circuits},
author={Zhizhen Chen and Daniel Mingyi Israel and Guy Van den Broeck and Zhe Zeng},
booktitle={The Ninth Workshop on Tractable Probabilistic Models},
year={2026},
url={https://openreview.net/forum?id=orwlLyK403}
}

ProbMoE: Differentiable Probabilistic Routing for Mixture-of-Experts

Heng Zhao, Zilei Shao, Guy Van den Broeck, Zhe Zeng

International Conference on Machine Learning (ICML) , 2026

PDF Code

@article{zhao2026probmoe,
  title   = {ProbMoE: Differentiable Probabilistic Routing for Mixture-of-Experts},
  author  = {Zhao, Heng and Shao, Zilei and Van den Broeck, Guy and Zeng, Zhe},
  journal = {arXiv preprint arXiv:2606.01509},
  year    = {2026}
}

Automatic Layer Selection for Hallucination Detection

Xinpeng Wang, William X. Cao, Andrew Gordon Wilson, Zhe Zeng

International Conference on Machine Learning (ICML) , 2026

PDF Code

@article{wang2026automatic,
  title   = {Automatic Layer Selection for Hallucination Detection},
  author  = {Wang, Xinpeng and Cao, William X. and Wilson, Andrew Gordon and Zeng, Zhe},
  journal = {arXiv preprint arXiv:2605.26366},
  year    = {2026}
}

Learning Deep Generative Models under Hard Linear Equality Constraints

Ruoyan Li, Dipti Ranjan Sahu, Guy Van den Broeck, Zhe Zeng

Proceedings of Machine Learning Research (PMLR) , 2026

PDF

@inproceedings{li2026learning,
  title     = {Learning Deep Generative Models under Hard Linear Equality Constraints},
  author    = {Li, Ruoyan and Sahu, Dipti Ranjan and Van den Broeck, Guy and Zeng, Zhe},
  booktitle = {Proceedings of Machine Learning Research (PMLR)},
  year      = {2026}
}