Philosophy: My teaching integrates traditional lectures with evidence-based instructional practices, multimedia, equipping students with the technical skills and critical thinking necessary for their careers while preparing them to address contemporary challenges at local, national, and global levels.
MLforMSE: Machine Learning for Chemical Science / Materials Science and Engineering
The course content for MLforMSE typically combines foundational machine learning techniques with domain-specific applications relevant to materials discovery, design, and characterization. This course typically covers the following topics:
🧠 Machine Learning Basics
🧪 Materials Informatics
🗂️ Data Preprocessing
📈 Supervised Learning
🔍 Unsupervised Learning
🌐 Deep Learning
✅ Model Evaluation
🧬 Case Studies
⚖️ Ethics in AI for Materials
Detailed Contents are given below:
Chapter 1. Introduction to Machine Learning (ML)
Overview of ML types: supervised, unsupervised, reinforcement learning
Common ML algorithms (regression, classification, clustering)
Overfitting, bias-variance tradeoff
Tools: Python, scikit-learn, TensorFlow, or PyTorch
Chapter 2. Materials Informatics Fundamentals
Data-driven materials science: motivation and use cases
Materials Simulations: ML for DFT, AIMD, Classical MD, Force Fields or Potentials
Materials databases: Materials Project, OQMD, AFLOW, NOMAD, and others
Materials descriptors and feature engineering
Structure-property relationships
Chapter 3. Data Handling for Materials Science
Materials data types: crystalline structures, spectroscopy, images, etc.
Data cleaning, curation, normalization
Feature extraction: composition-based, structure-based, image-based features
Use of domain-specific libraries (e.g., pymatgen, matminer)
Chapter 4. Supervised Learning Applications
Predicting materials properties (e.g., bandgap, formation energy)
Regression models: Linear Regression, Random Forests, SVR, Neural Networks
Classification models: SVMs, Decision Trees, k-NN, Deep Learning
Chapter 5. Unsupervised Learning & Dimensionality Reduction
Clustering of materials (e.g., for phase discovery)
PCA, t-SNE, UMAP for visualization and feature reduction
Pattern recognition in materials datasets
Chapter 6. Deep Learning in Materials Science
Neural network architectures
Convolutional Neural Networks (for images, microstructures)
Graph Neural Networks (e.g., crystal graph networks)
Transfer learning for small datasets
Chapter 7. Model Evaluation and Uncertainty
Cross-validation, metrics (MAE, RMSE, R², accuracy)
Uncertainty quantification and calibration
Interpretability (e.g., SHAP, LIME)
Chapter 8. Case Studies and Applications
Accelerated materials discovery
Inverse design of materials
High-throughput screening
Data-driven design of alloys, polymers, battery materials, etc.
Chapter 9. Ethical, Legal, and Societal Considerations
Data bias and reproducibility
Open science and data sharing
Limitations of ML in scientific discovery
Chapter 10. Hands-On Project or Final Assignment
Real-world dataset analysis or ML-driven materials design challenge
References: