CV

Education

Peking University, Ph.D. Candidate in Integrative Life Sciences (Physics), 2023 - 2028 expected

Academy for Advanced Interdisciplinary Studies
Advisor: Prof. Si Wu
Focus: cognitively inspired AI, world models, latent action models, skill learning, and computational neuroscience

Beijing Normal University, B.S. in Computer Science and Technology, 2019 - 2023

Please see the Publications page for the full list. Selected work:

Structure Abstraction and Generalization in a Hippocampal-Entorhinal Inspired World Model, ICML 2026. Co-first author.
DiLA: Disentangled Latent Action World Models, ICML 2026. Co-first author.
To Learn or Not to Learn, That is the Question - A Feature-Task Dual Learning Model of Perceptual Learning, NeurIPS 2024.
A Differentiable Approach to Multi-scale Brain Modeling, ICML Workshop 2024.

Structure Abstraction and Generalization in a Hippocampal-Entorhinal Inspired World Model, Peking University, Jul. 2024 - May 2026

Co-led a brain-inspired world model for hierarchical structure-content disentanglement.
Developed an HPC-like state module and MEC-like transition module for content-preserving sensory processing and path integration.
Trained on unlabeled human-object interaction videos, achieving strong performance on prediction and structure-reuse tasks.

DiLA: Disentangled Latent Action World Models, Peking University, Oct. 2025 - May 2026

Developed a dual-branch latent action world model that separates dynamics-related actions from static sequence content.
Learned continuous, semantically structured latent actions that transfer across embodiments and support MPC-based robotic simulation experiments.

Temporal Abstract Skill and Hierarchical Policy Learning from Unlabeled Video Data, Peking University, Jan. 2026 - Present

Developing methods to learn reusable temporal skill representations and hierarchical policies from unlabeled video data.
Designing the algorithmic framework and empirical validation pipeline for skill discovery, temporal dynamics modeling, and cross-embodiment transfer.

Additional Cognitive AI and Computational Neuroscience Research, Peking University and Beijing Normal University, 2020 - Present

Built computational models of high-frequency oscillations in epilepsy in collaboration with clinical neurosurgeons.
Contributed to a feature-task dual learning model of perceptual learning, published at NeurIPS 2024.
Built model-fitting modules for BrainPy and contributed to multi-scale differentiable brain modeling.
Worked on cognitive and brain-inspired AI projects spanning Raven's Progressive Matrices, spatial cognition based path planning, and hybrid brain-computer interfaces.

Teaching Assistant, Applications of AI in Psychological Research, Peking University, Feb. - Jul. 2025. Instructed 7 hours of core lectures on complex AI methodologies.
Instructor, International Summer Forum for Top Psychology Students, Aug. 2025. Led a 2-hour programming workshop with real-time technical troubleshooting.
Instructor, 4th Neural Computational Modeling Camp, Jul. 2025. Delivered a 3-hour lecture on simplified neuron models and dynamics analysis.

Oral Presenter, Soochow University-Peking University Joint Symposium on Cognitive Science and Brain-Computer Interfaces, Apr. 2026. Talk: "DiLA: Disentangled Latent Action World Models".
Poster Presenter, Center for Quantitative Biology Annual Meeting, Peking University, Jan. 2025. Poster: "To Learn or Not to Learn, That is the Question - A Feature-Task Dual Learning Model of Perceptual Learning".
Selected Participant, Cold Spring Harbor Asia Summer School on Computational and Cognitive Neuroscience, Suzhou, Jun. - Jul. 2024.

Meritorious Winner, Mathematical Contest in Modeling (MCM), 2021
First Prize (Beijing), Contemporary Undergraduate Mathematical Contest in Modeling (CUMCM), 2020
Jingshi Scholarship (Top 10%-20%), Beijing Normal University, 2020 - 2022
Cyrus Tang Foundation Scholarship, Cyrus Tang Foundation, 2020 - 2023

Programming / ML: Python, PyTorch, BrainPy
Research Areas: World models, latent action learning, skill learning, embodied AI, computational neuroscience
Tools: Weights & Biases, Hugging Face, Git, Linux/Bash, tmux, AI-assisted development with Claude/Codex
Languages: Mandarin (native), English (CET-6: 589), Japanese (JLPT N2)