Yibo Zhang

PhD candidate in Chemistry, University of New Hampshire

Currently

Intesting in materials science and machine learning

Specialized in

Machine Learning, Deep Learning, Data mining, Linux, Web development, Big data, SQL, Data visualization

Professional Experience

Jan. 2024 - Present Northeast Materials Database (NEMAD): Enabling Discovery of High Transition Temperature Magnetic Compounds

Supervisor: Prof. Jiadong Zang
Developed a comprehensive database of 26,706 magnetic materials using GPTArticleExtractor workflow
Incorporated chemical composition, structural details, and magnetic properties for each entry
Built machine learning models to classify materials and predict Curie and Néel temperatures
Achieved 90% accuracy in classifying materials as ferromagnetic, antiferromagnetic, or non-magnetic
Developed regression models predicting Curie temperature with R² of 0.86 and MAE of 62K
Created regression models for Néel temperature prediction with R² of 0.85 and MAE of 32K
Screened Materials Project database, identifying 62 potential high Curie temperature (>500K) ferromagnetic candidates
Established a user-friendly website (www.nemad.org) to host and provide access to the NEMAD database

Jan. 2023 - Jan. 2024 GPTArticleExtractor: Automated Magnetic Materials Database Construction

Supervisor: Prof. Jiadong Zang
Developed an innovative workflow using large language models to extract key information from scientific literature
Created a comprehensive database of 2,035 magnetic materials from 22,120 articles in the Journal of Magnetism and Magnetic Materials
Extracted and verified key material properties including chemical composition, structure, and magnetic temperatures
Achieved an 83.2% accuracy rate in identifying relevant articles and extracting useful information
Implemented a vector database approach to enhance the relevance and quality of extracted information
Utilized GPT-3.5 and GPT-4 models for different stages of the extraction process, optimizing for both efficiency and accuracy
Analyzed the distribution of Curie and Néel temperatures across different materials and space groups
Created a publicly accessible database at MagneticMaterials.org for wider scientific community use

Jan. 2022 – Jan. 2023 Magnetic vector field reconstruction

Supervisor: Prof. Jiadong Zang
Generate Magnetic vector field training dataset with Julia
Use U-Net convolutional neural networks to learn projected image-magnetic structure relationships
Train model with GPU acceleration
Analyze the prediction accuracy
The U-Net model can give accurate magnetic structure predictions by inputting random projection images

Mar. 2021 – Jan. 2022 Using Generative Adversarial Network (GAN) to reconstruct Magnetic Property from projection images

Supervisor: Prof. Jiadong Zang
Use the projection algorithm to project the three-dimensional magnetic structure into two-dimensional pictures
Build an autoencoder to analyze input images in manifold space
Build a generator to predict 3D magnetic properties by learning from input images
Build a discriminator to distinguish between real and fake generated magnetic structures
Assemble the generator and discriminator to form a Generative Adversarial Network (GAN) model
Let two models learn against each other to improve their accuracy
Work with Computer clusters to accelerate training time
GAN models can predict magnetic structure by learning from projected images

Nov. 2018 - Oct. 2020 Computational work on the study of Metal organic framework

Supervisor: Prof. Craig Chapman
Maintain a Linux server for computational work.
Use python script to analyze data
Use Jupyter notebook and python to manipulate the chemistry structure and generate input file
Remote manipulation of supercomputers
Density functional theory calculation to study the property of molecule

Tensroflow, PyTorch, Numpy, Pandas, Scikit-learn, Microsoft Machine Learning Studio
Linux, Bash script, Docker
D3, Matplotlib, Gnuplot
SQL, PySpark, Scala/Spark, OpenRefine, Tableau
AWS, GCP
Javascript, HTML, CSS,

Education

Jan. 2022-Now Georgia Institute of Technology, Online.

Computer science
Master’s Degree

Sep. 2018-Now University of New Hampshire, Durham.

Computational Chemistry
PhD’s Degree

Aug. 2015-May 2017 Stony Brook University, New York

Materials Science and Engineering
Master’s Degree

Sep. 2011-Jul. 2015 Zhengzhou University, China

Materials Chemistry, College of Materials Sciences and Engineering
Bachelor’s Degree

Work experience

Aug. 2019 - Now

Research assistant - University of New Hampshire

Aug. 2018 – Aug. 2019

Teaching assistant - University of New Hampshire

Oct. 2015 - Jan. 2017

Research assistant - Stony Brook University

Publications

Northeast Materials Database (NEMAD): Enabling Discovery of High Transition Temperature Magnetic Compounds

Itani, S., Zhang, Y., & Zang, J. arXiv preprint arXiv:2409.15675 (2024)

GPTArticleExtractor: An automated workflow for magnetic material database construction

Zhang, Y., Itani, S., Khanal, K., Okyere, E., Smith, G., Takahashi, K., & Zang, J. Journal of Magnetism and Magnetic Materials, 597, 172001 (2024)

Three-dimensional magnetization reconstruction from electron optical phase images with physical constraints

Lyu, B., Zhao, S., Zhang, Y., Wang, W., Zheng, F., Dunin-Borkowski, R. E., Zang, J., & Du, H. Science China Physics, Mechanics & Astronomy, 67(11), 1-11 (2024)

MagNet: machine learning enhanced three-dimensional magnetic reconstruction

Lyu, B., Zhao, S., Zhang, Y., Wang, W., Du, H., & Zang, J. arXiv preprint arXiv:2210.03066 (2022)

Conferences

October 2, 2024 IEEE AtC-AtG Magnetics Conference 2024

Oral presentation: “Comprehensive Database of Magnetic Materials Using AI-Driven Methodologies”

October 14, 2022 2022 Fall meeting of the New England sections (NES) of APS

Poster presentation: “MagNet: machine learning enhanced three-dimensional magnetic reconstruction”