Ze Chang

Led Multidisciplinary Project:

• Acted as project leader, fostering collaborations with various departments at Eindhoven University of Technology (TU/e), including multiscale lab, polymer lab, applied physics lab, and AMOLF.

Innovative Metamaterial Design:

• Introduced a Genetic Algorithm Optimized Approach for Metamaterial Design, achieving 8-fold increase in efficiency.

Advanced Optimization Techniques:

• Integrated Reinforcement Learning (Q-learning) into metamaterial design optimization, supported by experimental validation.

Machine Learning Model Development:

• Developed a U-net Convolutional Neural Network (CNN) to analyze correlation between crack morphology and microstructures, achieving Intersection over Union (IoU) score of 0.85 for predicting crack patterns.

Complex Problem Solving:

• Created Advanced Ensemble Learning (AEL) framework integrating Principal Component Analysis (PCA), Gaussian processes, and Light Gradient Boosting Machine (LGBM) algorithms to accurately predict Early-Age Shrinkage (EAS) evolution of concrete.

Data Science Solutions Development:

• Implemented a Residual Convolutional Neural Network (Res-Net) to predict elastic modulus and hardness of cement paste from Backscattered Electron (BSE) images, achieving R² values of 0.85 and 0.88, respectively, surpassing traditional models.

Model Interpretability:

• Utilized SHapley Additive exPlanations (SHAP) to identify key factors affecting concrete creep, ensuring transparency and interpretability of machine learning predictions.