Muhan Zhang
助理教授
Researches graph machine learning, graph neural networks and graph foundation models, and efficient LLM architectures and reasoning.
Muhan Zhang's lab at Peking works on graph learning, generative & foundation models and efficient training & inference. 77 papers since 2023, 61 at top venues, mostly NeurIPS and ICML. 9 PhD students. 4 alumni. Applications close .
At a glance
- papers, 2023–2026
- 20/ year
- at T1 venues
- 79%
- PhD students
- 9
- MS students
- 0
- alumni listed
- 4
Papers by yearfrom the lab homepage, 2023–2026
- 25 papers2026 · T1 19 · collecting
- 23 papers2025 · T1 17
- 19 papers2024 · T1 16
- 10 papers2023 · T1 9
T1 venuesOtherVenues: NeurIPS, ICML, ICLR, ACL, KDD
Research focuspaper titles that use each topic's words
- Graph learning35 papers
- graph26 papers
- Generative & foundation models16 papers
- Efficient training & inference10 papers
- researches graph0 papers
Paper list
From the lab homepage.
202625 papers
- NeurIPS (T1)CanvasMAR: Improving Masked Autoregressive Video Prediction With Canvas
- ICML (T1)Coevolutionary Continuous Discrete Diffusion: Make Your Diffusion Language Model a Latent Reasoner
- NeurIPS (T1)Cross Flow: One-Step Generation Across Latent and Pixel Spaces
- ICML (T1)Expressive Graph Neural Networks via Equivariant Use of Noise
- NeurIPS (T1)FlashMol: High-Quality Molecule Generation in as Few as Four Steps
- KDD-26GREPO: A Benchmark for Graph Neural Networks on Repository-Level Bug Localization
- KDD (T1)GRIP: In-Parameter Graph Reasoning through Fine-Tuning Large Language Models
- COLM-26HISA: Efficient Hierarchical Indexing for Fine-Grained Sparse Attention
- ACL (T1)JurisBench: A Deep Benchmark for Assessing Large Language Models in Professional Legal Practice
- ACL (T1)Knowledge is Not Enough: Injecting RL Skills for Continual Adaptation
- ACL-26Law in Silico: Simulating Legal Society with LLM-Based Agents
- ICML (T1)LIFT: A Novel Framework for Enhancing Long-Context Understanding of LLMs via Long Input Fine-Tuning
202523 papers
- DAC (T1)3D-SubG: A 3D Stacked Hybrid Processing Near/In-Memory Accelerator for Subgraph GNNs
- Nature Communications (T1)A data-driven group retrosynthesis planning model inspired by neurosymbolic programming
- Artificial Intelligence ReviewAdaptive graph diffusion networks: compact and expressive GNNs with large receptive fields
- NeurIPS (T1)Beyond Single-Task: Robust Multi-Task Length Generalization for LLMs
- ICML (T1)CLOVER: Cross-Layer Orthogonal Vectors Pruning and Fine-Tuning
- LoG-25Efficient neural common neighbor for temporal graph link prediction
- Frontiers of Computer ScienceExact Acceleration of Subgraph Graph Neural Networks by Eliminating Computation Redundancy
- LoG-25FLORA: GNNs as Predictors of Agentic Workflow Performances
- ICML (T1)Geometric Representation Condition Improves Equivariant Molecule Generation
- ICLR (T1)GOFA: A Generative One-For-All Model for Joint Graph Language Modeling
- ICML (T1)Griffin: Towards a Graph-Centric Relational Database Foundation Model
- EMNLP (T1)HD-PiSSA: High-Rank Distributed Orthogonal Adaptation
202419 papers
- ICML (T1)(ICML-24), 2024
- NeurIPS (T1)4DBInfer: A 4D Benchmarking Toolbox for Graph-Centric Predictive Modeling on Relational DBs
- Journal of Applied and Computational TopologyAmbiguities in neural-network-based hyperedge prediction
- KDD (T1)An Efficient Subgraph GNN with Provable Substructure Counting Power
- ICML (T1)An Empirical Study of Realized GNN Expressiveness
- ICML (T1)Case-Based or Rule-Based: How Do Transformers Do the Math?
- WWW (T1)Graph Contrastive Learning Meets Graph Meta Learning: A Unified Method for Few-shot Node Tasks
- ACL (T1)LooGLE: Can Long-Context Language Models Understand Long Contexts?
- NeurIPS (T1)Mars: Situated Inductive Reasoning in an Open-World Environment
- ICLR (T1)Neural Common Neighbor with Completion for Link Prediction
- ICLR (T1)On the Stability of Expressive Positional Encodings for Graph Neural Networks
- ICLR (T1)One For All: Towards Training One Graph Model For All Classification Tasks
202310 papers
- ICLR (T1)Boosting the Cycle Counting Power of Graph Neural Networks with I2-GNNs
- ICLR (T1)CktGNN: Circuit Graph Neural Network for Electronic Design Automation
- NeurIPS (T1)Distance-Restricted Folklore Weisfeiler-Lehman GNNs with Provable Cycle Counting Power
- NeurIPS (T1)Extending the Design Space of Graph Neural Networks by Rethinking Folklore Weisfeiler-Lehman
- NeurIPS (T1)Facilitating Graph Neural Networks with Random Walk on Simplicial Complexes
- ICML (T1)From Relational Pooling to Subgraph GNNs: A Universal Framework for More Expressive Graph Neural Networks
- NeurIPS (T1)Is Distance Matrix Enough for Geometric Deep Learning?
- ASP-DAC-23MacroRank: Ranking Macro Placement Solutions Leveraging Translation Equivariancy
- NeurIPS (T1)MAG-GNN: Reinforcement Learning Boosted Graph Neural Network
- VLDB (T1)SUREL+: Moving from Walks to Sets for Scalable Subgraph-based Graph Representation Learning
Where alumni went4 listed
Apply to PekingFall 2027
Computer Science PhD (international applicants)
- Round
- Main
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USCMU
Gauri Joshi
- Learning theory & optimization
- SGNTU
Jun Luo
- Signal processing & information theory
- CNTsinghua
Dan Pei
- Time series, anomaly & spatio-temporal data
- USCMU
Larry Wasserman
- Learning theory & optimization
- USStanford
Sherri Rose
- Learning theory & optimization
- USMIT
Guy Bresler
- Learning theory & optimization
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
Pekingcloses