|
Chuqin (Allen) Geng I am a Computer Science Ph.D. student at McGill University and Mila - Quebec AI Institute. I am currently a visiting scholar at the University of Toronto (U of T), working with Prof. Xujie Si. Prior to that, I earned my M.Sc. in Computer Science at Georgia Tech, and my B.Sc. in Math & Statistics at the University of Toronto. |
|
ResearchI am broadly interested in neural network interpretability and neuro-symbolic methods, with the goal of building transparent and reliable AI systems. My work spans interpretable and reliable learning for graph, vision, and language models, including applications in AI for science, as well as neural network robustness and formal verification. I currently study the mechanistic interpretability of LLMs and AI agents, tracing the representations and circuits behind their reasoning, planning, and decisions. |
Selected Papers |
|
Are We Recovering Mechanisms? Objective-Level Recovery Gaps in Mechanistic Interpretability [paper]
Chuqin Geng, Li Zhang, Haolin Ye, Mark Zhang, Luke Zhang, Xujie Si In submission |
|
Rethinking Circuit Evaluation: Do Circuits Explain Model Errors? [paper]
Li Zhang*, Chuqin Geng*, Mark Zhang, Chen Yang, Luke Zhang, Haolin Ye, Xujie Si In submission |
|
VisionLogic: Discovering and Grounding Decision-Relevant Visual Concepts [paper]
Chuqin Geng, Yuhe Jiang, Li Zhang, Zhaoyue Wang, Haolin Ye, Mark Zhang, Jingkai Xu, Xujie Si In submission |
|
Beyond Message Passing: A Symbolic Alternative for Expressive and Interpretable Graph Learning [paper]
Chuqin Geng, Li Zhang, Haolin Ye, Ziyu Zhao, Yuhe Jiang, Tara Saba, Xinyu Wang, Xujie Si In submission |
|
CROSSREF-VQA: Benchmarking Cross-modal Reference Resolution in Structured Graphics
Yuhe Jiang, Youyuan Zhang, Hangrui Bi, Chuqin Geng, Xujie Si In submission |
|
Neural Proposals, Symbolic Guarantees: Neuro-Symbolic Graph Generative Modeling [paper]
Chuqin Geng, Li Zhang, Mark Zhang, Haolin Ye, Ziyu Zhao, Xujie Si NeurIPS 2026 |
|
DreamProver: Evolving Transferable Lemma Libraries via a Wake-Sleep Theorem-Proving Agent [paper]
Youyuan Zhang*, Jialiang Sun*, Hangrui Bi, Chuqin Geng, Wenjie Ma, Zhaoyu Li, Xujie Si COLM 2026 |
|
LogicXGNN: Grounded Logical Rules for Explaining Graph Neural Networks [paper]
Chuqin Geng, Ziyu Zhao, Zhaoyue Wang, Haolin Ye, Yuhe Jiang, Xujie Si ICLR 2026 (Top ~2%) |
|
NEUROLOGIC: From Neural Representations to Interpretable Logic Rules [paper]
Chuqin Geng, Anqi Xing, Li Zhang, Ziyu Zhao, Yuhe Jiang, Xujie Si UCRL @ ICLR 2026 |
|
Learning Minimal Neural Specifications [paper]
Chuqin Geng, Zhaoyue Wang, Haolin Ye, Xujie Si NeuS 2025 (Oral) |
|
Towards Robust Saliency Maps [paper]
Van Nham Le*, Chuqin Geng*, Xujie Si, Arie Gurfinkel ACML 2024 |
|
Towards Reliable Neural Specifications [paper]
Chuqin Geng*, Van Nham Le*, Xiaojie Xu, Zhaoyue Wang, Arie Gurfinkel, Xujie Si ICML 2023 (Oral) |
|
Identifying Different Student Clusters in Functional Programming Assignments: From Quick Learners to Struggling Students [paper]
Chuqin Geng, Wenwen Xu, Yingjie Xu, Brigitte Pientka, Xujie Si SIGCSE TS 2023 |
|
TorchProbe: Fuzzing Dynamic Deep Learning Compilers [paper]
Qidong Su, Chuqin Geng, Gennady Pekhimenko, Xujie Si APLAS 2023 |
|
Scalar Invariant Networks with Zero Bias [paper]
Chuqin Geng, Xiaojie Xu, Haolin Ye, Xujie Si NeuReps @ NeurIPS 2023 |
|
Novice Type Error Diagnosis with Natural Language Models [paper]
Chuqin Geng, Haolin Ye, Yixuan Li, Tianyu Han, Brigitte Pientka, Xujie Si APLAS 2022 |
Academic ServiceI contribute to academic service through conference reviewing and student selection:
|
MiscellaneousOutside of research, I spend a lot of time outdoors, hiking and fishing whenever I get the chance. Being out in nature is when I do my best thinking, about how humans, animals, and the rest of the natural world all fit together, and how intelligence came to be in the first place. I also enjoy reading about paleontology, geology, and evolutionary biology in my spare time. |