Mr. Guangxuan Song | Neural Networks | Best Researcher Award

Ph.D. Candidate at University of Science and Technology Beijing, China

Mr. Guangxuan Song, is an accomplished Ph.D. student at the University of Science and Technology Beijing (USTB), specializing in knowledge graphs and graph neural networks applied to scientific data. His research focuses on applying advanced AI techniques to complex, noisy real-world datasets, particularly in materials science and corrosion prediction. He is proficient in Python, PyTorch, Torch Geometric, and LangChain, and also skilled in C++ and embedded systems. With a solid background in both theoretical and applied research, Mr. Song has developed web platforms, published cutting-edge papers, and participated in national projects like China’s Environmental Corrosion Platform. He is recognized for his leadership and innovation, having won national awards, including the Baidu Lingjing Developer title and multiple scholarships. Outside academia, he is passionate about photography, running, and mentorship. Mr. Song exemplifies a new generation of interdisciplinary researchers blending science, data, and innovation.

Professional Profile

Google Scholar

Education

Guangxuan Song began his academic journey at the University of Science and Technology Beijing (USTB), earning his Bachelor’s degree in Automation with distinction. During his undergraduate studies, he demonstrated exceptional academic performance and leadership, earning the National Scholarship in 2017 and recognition as the Beijing Outstanding Student in 2019. He continued at USTB for his Ph.D. in Control, focusing on cutting-edge AI research in scientific knowledge systems. His Ph.D. studies emphasized data-driven solutions for complex challenges in materials science and corrosion research, integrating machine learning, knowledge graph embeddings, and uncertainty modeling. Throughout his academic tenure, he participated in national science projects and received the National Scholarship for Ph.D. students in 2022. His education not only emphasized academic rigor but also highlighted leadership, public speaking, and collaborative research, evidenced by roles like Vice President of the Youth League Committee and other student leadership positions.

Work Experience

Guangxuan Song has been deeply involved in national and institutional research projects in China. He contributed to the National Environmental Corrosion Platform of China and the “Belt and Road” Corrosion Big Data Platform, building data analysis frameworks and web platforms supporting scientific infrastructure. He has also played an integral role in projects for the State Grid Corporation of China, using AI and data mining to understand atmospheric corrosion in power grid materials. Mr. Song has led efforts in developing 3 major web platforms, applied for 6 invention patents, and earned 7 software copyrights. He has proposed novel AI models (MPNNs, TSNet, numerical-semantic embeddings), setting new benchmarks in noisy data modeling and metal property prediction. His professional experience is not limited to technical work—he also led interdisciplinary collaborations, mentored peers, and integrated technical outcomes with public policy and industrial needs, proving his capability in both academic and practical environments.

🏆 Awards & Honors

Guangxuan Song has been widely recognized for academic excellence and innovative research. His accolades include the National Scholarship (2017, 2022), Dean’s Medal (2019), and Beijing Outstanding Graduate (2020). He received First Prize in the iCAN International Entrepreneurship Competition and First Prize in the “Internet+” Innovation Competition for his groundbreaking technological innovations. USTB honored him as an Inspirational Figure and Science and Technology Star for his leadership and impact. He was also selected as a Baidu Lingjing Developer in 2023, acknowledging his expertise in large language model applications. As a Huawei Student Developer and Daxing Half Marathon finisher, Mr. Song exemplifies balance in intellectual, physical, and creative domains. His consistent excellence has made him a leading example of China’s next-generation technology scholars—bridging scientific inquiry, innovation, and real-world application.

Research Focus

Mr. Song’s research is centered on the intersection of graph neural networks, knowledge graphs, and scientific data modeling. He is particularly focused on the interpretability and uncertainty quantification of machine learning models under noisy and heteroscedastic conditions, often present in real-world scientific datasets. His major contributions include:

A learnable MPNN-based PageRank for author ranking in AI networks
Fusion of semantic knowledge with structured numerical data for metal property prediction
The TSNet model for decoupling noise in heteroscedastic environments, enabling better data quality and active learning
He is passionate about creating AI models that are not only accurate but also transparent, interpretable, and applicable to practical challenges in materials science, corrosion engineering, and scientific discovery. His work addresses the urgent need for smarter data pipelines and analytical tools in science, with potential applications in national infrastructure, defense, and industrial R&D.

🧠 Skills

Guangxuan Song is a highly technical and versatile researcher with advanced skills in:

Python Programming (expert in PyTorch, Torch Geometric, LangChain)
C++ Development for microcontrollers and embedded systems
Data Science: Uncertainty estimation, heteroscedastic modeling, multivariate analysis
Machine Learning: Graph Neural Networks (GNNs), message-passing models, interpretability
Web Development: Scientific data platforms and UI/UX for research tools
Software Engineering: Holds 6 patents and 7 software copyrights
Research Tools: GPT-3.5 for scientific data pipelines, LangChain for literature parsing
Project Leadership: Guided cross-functional teams on national research projects
He also demonstrates exceptional communication, strategic planning, and collaborative leadership, supported by his experience as a student leader and mentor. His hobbies in photography and marathon running reflect discipline, creativity, and endurance—traits that further enrich his academic and professional life.

Conclusion

Mr. Guangxuan Song embodies the qualities of an emerging top-tier researcher: innovative, cross-disciplinary, and dedicated to solving real-world problems with advanced data-driven techniques. His balance of deep technical skills, pioneering research, national-level impact, and entrepreneurial achievements make him highly suitable for the Research for Best Researcher Award. With further global exposure and continued publishing excellence, he is positioned to become a future leader in AI-driven scientific discovery.

Publications to Noted

Cross-category prediction of corrosion inhibitor performance based on molecular graph structures via a three-level message passing neural network model
🧑‍🔬 Authors: J. Dai, D. Fu, G. Song, L. Ma, X. Guo, A. Mol, I. Cole, D. Zhang
📊 Citations: 17
📅 Year: 2022
📰 Journal: Corrosion Science 209, 110780

From Knowledge Graph Development to Serving Industrial Knowledge Automation: A Review
🧑‍🔬 Authors: G. Song, D. Fu, D. Zhang
📊 Citations: 6
📅 Year: 2022
📰 Conference: 2022 41st Chinese Control Conference (CCC), pp. 4219–4226

Bridging the Semantic-Numerical Gap: A Numerical Reasoning Method of Cross-modal Knowledge Graph for Material Property Prediction
🧑‍🔬 Authors: G. Song, D. Fu, Z. Qiu, Z. Yang, J. Dai, L. Ma, D. Zhang
📊 Citations: 2
📅 Year: 2023
📰 Repository: arXiv preprint arXiv:2312.09744

A Named Entity Extraction Method for Commonly Used Steel Knowledge Graph
🧑‍🔬 Authors: Z. Ma, L. Ma, D. Fu, G. Song, D. Zhang
📊 Citations: 2
📅 Year: 2022
📰 Conference: 2021 Chinese Intelligent Systems Conference: Volume III, pp. 724–732

Corrosion Resistant Performance Prediction in High-Entropy Alloys: A Framework for Model, Interpretation and Multi-Dimensional Visualization
🧑‍🔬 Authors: G. Song, D. Fu, W. Chang, Z. Fu, L. Ma, D. Zhang
📊 Citations: Not available (recent publication)
📅 Year: 2025
📰 Journal: Corrosion Science, 113105

A Message Passing Neural Network Framework with Learnable PageRank for Author Impact Assessment
🧑‍🔬 Authors: S. Guangxuan, D. Fu, X. Wu
📊 Citations: Not available (recent publication)
📅 Year: 2025
📰 Journal: Advances in Electrical & Computer Engineering 25 (1)

Taylor-Sensus Network: Embracing Noise to Enlighten Uncertainty for Scientific Data
🧑‍🔬 Authors: G. Song, D. Fu, Z. Qiu, J. Meng, D. Zhang
📊 Citations: Not available (recent preprint)
📅 Year: 2024
📰 Repository: arXiv preprint arXiv:2409.07942

A Data-Driven Hybrid Machine Learning and Transfer Learning Algorithm for Long Consecutive Atmospheric Environment Missing Data Imputation
🧑‍🔬 Authors: H. Meng, L. Shao, D. Fu, G. Song, J. Meng
📊 Citations: Not available (recent publication)
📅 Year: 2024
📰 Conference: 2024 43rd Chinese Control Conference (CCC), pp. 6728–6733

Extraction of Key Environmental Factors and Construction of the Data-Driven Model for Atmospheric Aging of Polyurethane Coatings
🧑‍🔬 Authors: Z. Bai, D. Fu, G. Song, L. Shao, D. Zhang
📊 Citations: Not available
📅 Year: 2023
📰 Conference: 2023 42nd Chinese Control Conference (CCC), pp. 6713–6718

Guangxuan Song | Neural Networks | Best Researcher Award