Jiachen Jiang

Biography

I am advised by Prof. Zhihui Zhu at The Ohio State University. My work spans various directions:

• Layerwise representation analysis in deep networks. Characterizing how representations evolve across layers to reveal what each layer learns and how it shapes generalization.
• In-context learning in large language models. Understanding the internal mechanisms of in-context learning and extending it to more general settings.
• LLM-driven scientific discovery. Designing principled algorithms that turn LLMs into search engines for scientific codes.
• Multimodal LLMs and image restoration. Improving vision–language alignment and building content- and task-aware models for unified low-level vision tasks.

Feel free to reach out to me by email—I am always happy to chat about research and collaborations!

Education

• The Ohio State University (OSU) Columbus, OH
  Doctor of Philosophy in Computer Science and Engineering 2023 – Present
• University of Michigan (UMich) Ann Arbor, MI
  Master of Science in Electrical & Computer Engineering 2020 – 2023
• Beihang University (BUAA) Beijing, China
  Bachelor of Science in Electrical Engineering 2016 – 2020

Experience

• Research Intern May 2026 – Aug. 2026
  @ Microsoft, Redmond, WA
• Research Intern Jun. 2024 – Sep. 2024
  @ Microsoft, Redmond, WA
• Research Assistant Aug. 2023 – Present
  @ The Ohio State University, Columbus, OH

Selected Publications

(★ indicates first-author work)

1. DeltaEvolve: Accelerating Scientific Discovery through Momentum-Driven Evolution
   Jiachen Jiang, Tianyu Ding, Zhihui Zhu
   International Conference on Machine Learning (ICML), 2026 AC rate = TBD
   pdf arXiv
2. Improving Visual Discriminability of CLIP for Training-Free Open-Vocabulary Semantic Segmentation
   Jinxin Zhou, Jiachen Jiang, Zhihui Zhu
   Transactions on Machine Learning Research (TMLR), 2026
   pdf arXiv
3. Understanding Task Vectors in In-Context Learning: Emergence, Functionality, and Limitations
   Yuxin Dong, Jiachen Jiang, Zhihui Zhu, Xia Ning
   International Conference on Learning Representations (ICLR), 2026 AC rate = 27.4%
   pdf arXiv
4. Learning to Adapt: In-Context Learning Beyond Stationarity
   Zhen Qin, Jiachen Jiang, Zhihui Zhu
   International Conference on Learning Representations (ICLR), 2026 AC rate = 27.4%
   pdf arXiv
5. From Compression to Expansion: A Layerwise Analysis of In-Context Learning
   Jiachen Jiang, Yuxin Dong, Jinxin Zhou, Zhihui Zhu
   International Conference on Learning Representations (ICLR) DeLTa Workshop, 2026
   pdf arXiv project page
6. On the Convergence of Gradient Descent on Learning Transformers with Residual Connections
   Zhen Qin, Jinxin Zhou, Jiachen Jiang, Zhihui Zhu
   IEEE Signal Processing Letters (SPL), 2026
   pdf arXiv IEEE
7. ProCrop: Learning Aesthetic Image Cropping from Professional Compositions
   Ke Zhang, Tianyu Ding, Jiachen Jiang, Tianyi Chen, Ilya Zharkov, Vishal M. Patel, Luming Liang
   AAAI Conference on Artificial Intelligence (AAAI), 2026 AC rate = 17.6%
   pdf arXiv project page
8. Analyzing Fine-Grained Alignment and Enhancing Vision Understanding in Multimodal Language Models
   Jiachen Jiang, Jinxin Zhou, Bo Peng, Xia Ning, Zhihui Zhu
   Neural Information Processing Systems (NeurIPS), 2025 AC rate = 24.5%
   pdf arXiv
9. Can Reasoning LLMs Eliminate Conformity in Multi-Agent Systems? Best Paper Award
   Alina Liu, Christopher Hill, Jiachen Jiang, Zhihui Zhu
   IEEE International Conference on Data Mining Workshops (ICDMW), 2025
   pdf
10. Tracing Representation Progression: Analyzing and Enhancing Layer-Wise Similarity
    Jiachen Jiang, Jinxin Zhou, Zhihui Zhu
    International Conference on Learning Representations (ICLR), 2025 AC rate = 32%
    pdf arXiv
11. Are All Layers Created Equal: A Neural Collapse Perspective
    Jinxin Zhou, Jiachen Jiang, Zhihui Zhu
    Conference on Parsimony and Learning (CPAL), 2025
    pdf
12. Generalized Neural Collapse for a Large Number of Classes
    Jiachen Jiang, Jinxin Zhou, Peng Wang, Qing Qu, Dustin Mixon, Chong You, Zhihui Zhu
    International Conference on Machine Learning (ICML), 2024 AC rate = 27.5%
    pdf arXiv
13. DREAM: Diffusion Rectification and Estimation-Adaptive Models
    Jinxin Zhou, Tianyu Ding, Tianyi Chen, Jiachen Jiang, Ilya Zharkov, Zhihui Zhu, Luming Liang
    Computer Vision and Pattern Recognition (CVPR), 2024 AC rate = 23.6%
    pdf code arXiv project page

Selected Preprints

1. SMCEvolve: Principled Scientific Discovery via Sequential Monte Carlo Evolution
   Jiachen Jiang, Huminhao Zhu, Zhihui Zhu
   arXiv preprint, 2026
   pdf arXiv
2. In-Context Learning for Non-Stationary MIMO Equalization
   Jiachen Jiang, Zhen Qin, Zhihui Zhu
   arXiv preprint, 2025
   pdf arXiv
3. From Emergence to Control: Probing and Modulating Self-Reflection in Language Models
   Xudong Zhu, Jiachen Jiang, Mohammad Mahdi Khalili, Zhihui Zhu
   arXiv preprint, 2025
   pdf arXiv
4. Cat-AIR: Content and Task-Aware All-in-One Image Restoration
   Jiachen Jiang, Tianyu Ding, Ke Zhang, Jinxin Zhou, Tianyi Chen, Ilya Zharkov, Zhihui Zhu, Luming Liang
   arXiv preprint, 2025
   pdf arXiv project page
5. The Efficiency Spectrum of Large Language Models: An Algorithmic Survey
   Tianyu Ding, Tianyi Chen, Haidong Zhu, Jiachen Jiang, Yiqi Zhong, Jinxin Zhou, Guangzhi Wang, Zhihui Zhu, Ilya Zharkov, Luming Liang
   arXiv preprint, 2023
   pdf arXiv

Visitors