Projects Experience

Generative Characterization of Oil-Gas Reservoirs, Chinese Academy of Science.

2024/09 - 2025/03

• Proposed a Discrete Neural Operator with Adaptive Sampling for transient Darcy flow surrogate modeling (paper under review, first author)
• Achieved 24.3% lower error than Residual Attention U-Net; adaptive sampling improved generalization by 35% over random sampling

Generative Large Model for Hydrocarbon Sweet Spots, Chinese Academy of Science.

2023/11 - 2024/03

• Developed dataset augmentation via sublinear expectation theory to enable small-sample training
• Filed patent: A Small-Sample Learning Method Based on Mathematical Expectation

Intelligent Surrogate Model for Imbalanced Data, Qingdao Soft Control Company.

2023/09 - 2024/09

• Addressed class imbalance in rubber-tire manufacturing via synthetic oversampling and sublinear-expectation-based noise modeling.
• Improved classification accuracy by 13% using logistic regression under maximal distribution

Powered by Jekyll and Minimal Light theme.