SUBHANKAR MISHRA
ଶୁଭଙ୍କର ମିଶ୍ର
Faculty, School of Computer Sciences, NISER
ପାଠକ-ଏଫ, ସଂଗଣକ ବିଜ୍ଞାନ ବିଦ୍ୟାଳୟ, ନାଇଜର
Machine Learning for Materials Science
About
Machine Learning is transforming the area of materials science by aiding materials screening, discovery and design. ML techniques are already cutting down the time required for discovery of novel functional materials significantly, and this trend is going to increase going forward. Eventually a big part of materials science is going to be done in autonomous, self-driving labs, aided by ML models including generative AI.In order to prepare the next generation of researchers and science/technology experts in these emerging areas, this course will introduce the foundational concepts of Machine Learning relevant to materials science. No prior knowledge of ML is assumed. However, knowledge of basic Solid State Physics/Chemistry and Python scripting are required.
At the end of the course, students will be able to use ML in their area of study/research using available data and standard packages.
Instructors
- Prof. Prasenjit Sen, HRI
- Subhankar Mishra, NISER
Teaching Assistants
- Sourav Mal, PhD Candidate HRI
- Rishiraj Sahoo, PhD Student NISER
How to register?
- Target students: M. Sc, M. Tech. and Ph. D. Students
- Students interesting in attending the course have to register on the Anuvidhya website
- Help Manual for Registration of Online Course Link
- Deadline for registration: August 1, 2025.
Class hours
- Classes from August 11, 2025 to December 12, 2025
- Class timings: 4:30 - 5:30 PM, Mondays, Wednesdays and Fridays (with breaks during major festivals)
Syllabus
- Machine Learning Basics: Machine Learning Introduction, Types of Machine Learning (Input/Output) Linear Regression and Logistic Regression Decision Trees, Support Vector Machines (Linear SVM), Support Vector Machines (Soft SVM), Support Vector Machines (Non-Linear SVM), kNN, Loss functions, Gradient Descent Feature Engineering, Dataset split, Underfitting and Overfitting, Bias and Variance Regularization, Types of Machine Learning (Input/Output), k-Means, DBSCAN, PCA, tSNE, Neural Networks, Multilayer Perceptron Feedforward Neural Networks, Back-propagation, Deep Learning, Convolutional Neural Network.
- Applications to Materials Research: Materials databases, Downloading materials data, Handling materials data, Features representing crystalline materials, Feature extraction from materials data, 2D point cloud representation of materials, Preparing data for model training, Model training, Cross validation, Hyper-parameter optimization, Performance metrics of regression and classification models, Hands-on training on materials data, Hands-on examples of materials property predictions and materials classification, Advanced applications of ML in materials science: expository lecture(s).
Prerequisites
BSc/BS, BE/BTech (for students in Master degree), MSc/MS or ME/MTech (for students pursuing PhD) with at least one course on solid state physics/chemistry, and knowledge of python scripting. Familiarity with first-principles density functional theory calculations and/or experimental techniques is desirable, but not essential.Books
Some recommended books on Machine learning.- The Elements of Statistical Learning Link
- Mathematics for Machine Learning Link
- Introduction to Machine Learning with Python Link
- A course in Machine Learning Link