Jiaxuan Zou 邹嘉轩

Jiaxuan Zou 邹嘉轩

Deep learning researcher

About Me

I am an undergraduate student at the School of Mathematics and Statistics, Xi’an Jiaotong University. Currently, I am also a research intern at the Gaoling School of Artificial Intelligence, Renmin University of China, advised by Prof. Yong Liu.

Research Interests

  • Mechanistic Interpretability of Large Language Models
  • Deep Learning Theory and Optimization
  • AI for Scientific Discovery and Scientific Modeling

Main Research Experience

Gaoling School of Artificial Intelligence, Renmin University of China

Research Intern | Jun. 2025 – Present

  • Investigated latent reasoning mechanisms in large language models through the lens of mechanistic interpretability.
  • Developed a theoretical framework to characterize the exploration-execution trade-off in Latent Chain-of-Thought.
  • Proposed the Symbolic Index, a novel metric to measure the model’s decisional certainty during reasoning.
  • Proved the necessity of curriculum learning for achieving stable training in this context.
  • Work resulting from this research is currently under review at ICLR (incoming!).

School of Mathematics and Statistics, Xi’an Jiaotong University

Undergraduate Researcher | 2024 – Present

  • Developed a channel-independent multi-scale framework for seismic horizon picking, specifically designed to perform effectively with extremely limited labeled data. Manuscript under review at Geophysics.
  • Designed and implemented a novel U-Net architecture based on Mamba (a State Space Model) for automatic seismic interpretation, focusing on improving feature fusion and long-range dependency modeling. Manuscript under review at IEEE TGRS.

Publications

Conference Papers

  1. Sun, J., Ling, X., Zou, J., Kang, J., & Zhang, K. (2025). FreLE: Low-Frequency Spectral Bias in Neural Networks for Time-Series Tasks. In Proceedings of the IEEE International Conference on Data Mining (ICDM). (Accepted)

Journal Articles

  1. Zou, J., Liu, N., Zhang, T., He, J., Li, T., & Gao, J. Seismic Horizon Picking Using Channel-Independent Multi-Scale Network with Limited Training Samples. Submitted to Geophysics. (Under Review)
  2. Zhang, T., He, J., Liu, N., Zou, J., & Jiang, Y. Mamba-Driven and Feature-Fused U-Net for Automatic Seismic Horizon Interpretation. Submitted to IEEE Transactions on Geoscience and Remote Sensing (TGRS). (incoming!)