Open-World AI · Scientific Discovery

Guangyao Chen

DSAI Postdoctoral Fellow

Johns Hopkins University

I am a Postdoctoral Fellow in the Data Science and AI Institute (DSAI) at Johns Hopkins University. Previously, I was a Schmidt AI in Science Postdoctoral Fellow at the AI for Science Institute, Cornell University. I received my Ph.D. in Computer Science from Peking University and my B.S. in Computer Science from Wuhan University.

I am on the 2026–2027 job market for faculty and industry AI research positions.

Portrait of Guangyao Chen

Research

My research focuses on developing general-purpose AI systems that recognize scientific unknowns, formulate hypotheses grounded in existing evidence, and adaptively acquire new evidence to test and refine these hypotheses. By integrating these capabilities, I build closed-loop, verifiable systems for autonomous scientific discovery, with testbeds in scientific imaging, materials discovery, and molecular design.

My current research interests include:

  • Open-world learning for recognizing unknowns, detecting distribution shifts, and updating models as new evidence arrives (ARPL · OpenOOD · APR · MICM · EventAD).
  • Evidence-grounded agentic reasoning for coordinating specialized agents, preserving provenance, and producing auditable hypotheses and claims (AutoAgents · GroundingAgent · SymbolicDet).
  • Autonomous scientific discovery through evidence-seeking actions, experimental feedback, and iterative hypothesis validation (EMSeek · STEM2Crystal-Bench · PSCG-Net · GQP).

News

2026.07 Context-aware multimodal visual-RL paper accepted to IJCV.
2026.05 Interpretable peptide-screening paper published in Chemical Science.
2026.05 STEM crystal-reconstruction paper accepted as a SIGKDD 2026 Oral.
2026.04 EMSeek, our microscopy agent, published in Science Advances and featured by Cornell.
2026.02 Two source-free cross-domain few-shot learning papers accepted to CVPR 2026.
2025.11 Five visual-reasoning and few-shot learning papers accepted to AAAI 2026, including two orals.
2025.09 Crystal-graph materials-discovery paper published in JCIM.
2025.09 Few-shot class-incremental learning paper accepted to NeurIPS 2025.
2025.08 AI-driven particle-vision review published in Engineering.
2025.06 Symbolic visual-reasoning paper accepted to ICCV 2025.
2025.05 Three anomaly-detection and few-shot segmentation papers accepted to ICML 2025, including two spotlights.
2025.04 Language-model materials-intelligence paper published in Advanced Materials.
2025.03 Few-shot OOD-detection paper selected as a spotlight at the CVPR 2025 VAND Workshop.
2024.11 2024 CSIG Outstanding Ph.D. Thesis Award.
2024.09 Multimodal visual-RL paper accepted to NeurIPS 2024.
2024.07 Two few-shot and open-set learning papers accepted to ACM MM 2024, including one oral.
2024.04 Automatic agent-generation paper accepted to IJCAI 2024.
2023.12 Incremental novel-class discovery paper accepted to AAAI 2024.
2023.11 Invited talk on open-world learning at the Qingyuan Workshop (online).
2023.11 Invited talk on automatic agent generation at RLChina 2023, Suzhou.

Selected Publications

Complete list on Google Scholar

* Equal contribution; † corresponding author.

  1. Sci. Adv. 2026
    Bridging Electron Microscopy and Materials Analysis with an Autonomous Agentic Platform
    Guangyao Chen, Wenhao Yuan, and Fengqi You
    Science Advances, 2026
    MRS Oral Presentation
    Paper News Code
  2. TPAMI 2022
    Adversarial Reciprocal Points Learning for Open Set Recognition
    Guangyao Chen, Peixi Peng, Xiangqian Wang, and Yonghong Tian
    IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022
    Academician Shi Qingyun’s Outstanding Paper Award, Highly Cited Paper
    arXiv Paper Code
  3. IJCAI 2024
    AutoAgents: A Framework for Automatic Agent Generation
    Guangyao Chen*, Siwei Dong*, Yu Shu*, Ge Zhang, Sesay Jaward, and 3 more authors
    In The 33rd International Joint Conference on Artificial Intelligence, 2024
    Acceptance Rate: 15%
    arXiv Code
  4. ICCV 2021
    Amplitude-Phase Recombination: Rethinking Robustness of Convolutional Neural Networks in Frequency Domain
    Guangyao Chen, Peixi Peng, Li Ma, Jia Li, Lin Du, and 1 more author
    In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), Oct 2021
    arXiv Paper Code
  5. SIGKDD 2026
    From Noisy STEM to Crystal Structure: Evidence-Structure CoDiffusion under Composition Constraints
    Guangyao Chen, and Fengqi You
    In Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Oct 2026
    Oral Presentation, Acceptance Rate: < 6.1%
    Dataset Code
  6. ICML 2025
    When Every Millisecond Counts: Real-Time Anomaly Detection via the Multimodal Asynchronous Hybrid Network
    Dong Xiao*, Guangyao Chen*†, Peixi Peng, Yangru Huang, Yifan Zhao, and 2 more authors
    In Forty-second International Conference on Machine Learning, Oct 2025
    Spotlight Poster, Acceptance Rate: < 2.6%
    Paper
  7. Chem. Sci. 2026
    Rethinking Peptide Developability with Sequence-Only Models: Interpretable Screening of Microplastic-Binding Peptides with Gated Query Pooling
    Guangyao Chen, and Fengqi You
    Chemical Science, Oct 2026
    Front Cover
    Paper
  8. AAAI 2026
    Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning
    Liqin Luo*, Guangyao Chen*†, Xiawu Zheng, Yongxing Dai, Yixiong Zou, and 1 more author
    In The Fortieth AAAI Conference on Artificial Intelligence, Oct 2026
    arXiv Code
  9. ICCV 2025
    From Objects to Events: Unlocking Complex Visual Understanding in Object Detectors via LLM-guided Symbolic Reasoning
    Yuhui Zeng, Haoxiang Wu, Wenjie Nie, Guangyao Chen†, Xiawu Zheng, and 4 more authors
    In International Conference on Computer Vision, Oct 2025
    arXiv Code
  10. CVPR 2026
    Reclaiming Lost Text Layers for Source-Free Cross-Domain Few-Shot Learning
    Zhenyu Zhang, Guangyao Chen†, Yixiong Zou, Yuhua Li, and Ruixuan Li
    In IEEE/CVF Conference on Computer Vision and Pattern Recognition, Oct 2026
    arXiv Paper

Academic Service

Area Chair

NeurIPS 2025 & 2026 Workshops on AI for Science

ICML 2026 Workshop on AI for Science

Conference Reviewing

NeurIPS, ICLR, ICML, AAAI, IJCAI, CVPR, ICCV, ECCV, SIGKDD, and WACV.

Journal Reviewing

Science Advances, IEEE TPAMI, IJCV, TMLR, IEEE TIP, IEEE TNNLS, and Pattern Recognition.

Honors

2026

DSAI Postdoctoral Fellowship

Johns Hopkins University

2026

ICML Gold Reviewer

International Conference on Machine Learning

2024

Schmidt AI in Science Postdoctoral Fellowship

Schmidt Sciences

2024

CSIG Doctoral Dissertation Award

Top 10 nationwide · China Society of Image and Graphics

2024

Beijing Doctoral Dissertation Award

Top 1% in Beijing

2023

Outstanding Doctoral Dissertation Award

Top 1% · Peking University

2022

Emerging Technology Award

IEEE Standards Association

2022

Certificate of Appreciation

IEEE Standards Association

2021

National Scholarship

Top 1% · Peking University

2018

Outstanding Undergraduate Dissertation

Top 1% · Wuhan University