About
I am an undergraduate student at the School of the Gifted Young, University of Science and Technology of China (USTC), majoring in Information and Computing Science with a minor in Computer Science and Technology. I expect to graduate in June 2027.
My research interests lie at the intersection of 3D vision, computer graphics, and generative modeling, with a particular interest in learning structured representations of the 3D world from visual observations.
For my undergraduate theses, I am working with Prof. Ligang Liu at the School of Mathematical Sciences on my major thesis, and with Prof. Zhenbo Shi at the School of Computer Science and Technology on my Computer Science minor thesis.
Starting in September 2027, I will pursue my master's degree at the School of Artificial Intelligence and Data Science (SAIDS), USTC, under the supervision of Prof. Ke Xu, where I plan to focus on 3D vision and related topics in computer vision and computer graphics.
Previously, I was a research intern at Westlake University, where I worked on 3D reconstruction, flow matching, and visual perception. More broadly, I am interested in building learning systems that can understand, reconstruct, and eventually model the physical world.
我是陈博远，中国科学技术大学（USTC）少年班学院本科生，预计于2027年6月毕业。研究兴趣为三维视觉、计算机图形学与生成式建模；将于2027年9月进入中科大人工智能与数据科学学院攻读硕士学位。
Experience
Sep 2023 – Present
B.S. at School of the Gifted Young & CS minor at CS USTC
Selected Projects
Selected research and engineering projects by Boyuan Chen (Chen Boyuan).
3D Reconstruction & Flow Matching
Research Internship · Westlake University
Explored modern approaches to 3D reconstruction and generative modeling, with a focus on Flow Matching, VGGT, and learning-based 3D reconstruction. Studied how generative priors and geometry-aware representations can be used to recover 3D structure from visual observations, and summarized experiments and technical findings in a series of research reports.
Deep Learning for Stock Prediction & Trading
Built an end-to-end deep learning system for short-term return prediction and quantitative trading on the Chinese A-share market, covering data preprocessing, feature engineering, model training, time-series evaluation, realistic T+1 backtesting, and daily trading signal generation.
Experiments were conducted on approximately 7.47 million samples from 3,163 stocks, comparing MLP, LSTM, and DLinear models under multiple prediction targets. The system also includes transaction-cost analysis, market benchmarks, Top-K portfolio strategies, and a local Streamlit interface for generating daily buy, sell, and hold decisions.
Neural Operators for Atmospheric Pollution Modeling
Developed a Scientific Machine Learning framework for modeling atmospheric pollutant dispersion, comparing classical physical simulation with neural surrogate models.
The project evaluates Gaussian Plume models, Lagrangian Particle Tracking, U-Net, FNO, U-FNO, and Physics-FNO under single-source, multi-source, and out-of-distribution settings. A key part of the study investigated failure modes of neural surrogates trained on sparse particle fields and improved training stability through smoothed physical supervision and mass-normalized targets.
The final U-FNO model achieved spatial correlation above 0.90 on both single- and multi-source test scenarios and remained robust when generalized to unseen numbers of pollution sources.
Featured Writing · 博客精选
先有选择，再谈选择：保研复盘
保研结束以后回头看，我觉得自己这一年最大的收获，不是哪一场夏令营，也不是最后拿到了哪个老师的名额，而是终于想明白了一件很简单的事情：
在没有选择的时候，不要花太多时间纠结怎么选。先让自己拥有足够多的选择，再谈哪个最好。
这句话现在看起来很理所当然，但我整个保研期间其实一直在反着做。
Awards
- CCF Certified Software Professional - Junior Group Second Prize, 2019
- CCF Certified Software Professional - Junior Group First Prize, 2020
- National Mathematics League of high school,Zhejiang Division Second Prize, 2022
- 16th Mathematics competition of Chinese College Students(CMC), Anhui Division Second Prize, 2024
- Bronze Award, Outstanding Student Scholarship, 2024
- 17th Mathematics competition of Chinese College Students(CMC), Anhui Division Second Prize, 2025