MSE598 ML :: MatSE Illinois :: University of Illinois at Urbana-Champaign
Home
Online discussion forum
This class uses the CampusWire System for announcements, updates, and all communication. Please visit this page to access it.
Excused Absences
Excused absences may be requested by filling out the Excused Absences form.
For more information, please read the course syllabus.
Schedule
Recordings will be posted under this link.
Date
|
Reading Material
|
Topic/Description
|
Notes/Assignments
|
1/21 |
Anaconda, Article |
Course Outline, Computer Environment, Statistics |
|
1/23 |
ML online course, When Not To Use ML |
Overview: Supervised, Unsupervised, Reinforcement Learning |
|
1/28 |
Youtube Link, Youtube Link, ML online course |
Deployment, Hands-on: Introduction Python/Jupyter, numpy |
|
1/30 |
Youtube Link, Youtube Link, ML online course |
Random numbers, Hands-on: Linear Algebra and Statistics with Python |
|
2/4 |
Youtube Link |
Data: Data types, data sources, cleanup, preparation, curation |
|
2/6 |
FAIR, ML online course, ML online course, Youtube Link, Youtube Link |
Data curation; FAIR principles; Hands-on: pandas |
|
2/11 |
ML online course, Youtube Link |
Hands-on: matplotlib, Data analysis; Models: Linear Regression, Validation |
|
2/13 |
ML online course |
Hands-on and Models: Linear regression |
Upload first notebook |
2/18 |
Paper, ML online course, Youtube Link |
Hands-on and Hands-on and Models: Linear regression, Evaluation, Feature Engineering/Descriptor selection (LASSO), Tuning and Model selection |
|
2/20 |
ML online course |
Models: Classification |
|
2/25 |
Youtube Link, Youtube Link, Youtube Link |
Models: Classification, Training |
Andre on travel |
2/27 |
Google Document, Notebook Link |
Ridge Regression |
|
3/4 |
|
|
Andre on travel |
3/6 |
IBM Document, Youtube Link, Youtube Link, Youtube Link, Youtube Link, Notebook Link |
Normalization; Cross Validation; Models: Kernel Ridge Regression, Models: K nearest neighbors regression; Unsupervised Learning:Clustering, K means, DBSCAN
| Watch Video 1, Watch Video 2 |
3/11 |
Youtube Link, Nanohub Video, Nanohub Material |
Models: Principal Component Analysis, Non-negative matrix factorization |
Watch Video 1, Watch Video 2 |
3/13 |
NOMAD API description and video |
Data: Usage of databases (Web and API) |
Watch Video 1, Watch Video 2 |
3/18 |
|
Spring Break |
|
3/20 |
|
Spring Break |
|
3/25 |
|
Catchup slides, Notebook 1, Notebook 2 |
|
3/27 |
Youtube Link, Youtube Link, Paper, Youtube Link |
Compressed Sensing, Support Vector Machines |
Upload second notebook |
4/1 |
ML online course, Youtube Link, Paper |
Evaluation of Classifiers, Models: Gaussian Process Regression, Hands-on |
|
4/3 |
Youtube Link |
Hands-on and Models: Gaussian Process Regression; Active Learning via Bayesian Optimization |
|
4/8 |
ML online course |
Models: Active Learning, Decision Trees |
Andre possibly on travel |
4/10 |
ML online course |
Models and Hands-on: Trees and Ensemble Learning, Decision Trees, Forests, Boosting |
Andre possibly on travel |
4/15 |
ML online course, Youtube Link, IBM Link |
Models: Neural Networks and Deep learning; Hands-on: Trees and Forests |
|
4/17 |
Youtube Link, Youtube Link, Youtube Link, Youtube Link |
Models: Training of NNs via Backpropagation |
|
4/22 |
ML course, Youtube Link, atomagined GitHub |
Models: Convolutional NNs for Image analysis; Hands-on: Neural Networks |
|
4/24 |
Paper, Youtube Link, Paper |
Data: Atomistic Descriptors (SMILES, SOAP, Coulomb matrix, MBTR); Ontologies; Hands-On |
|
4/29 |
Youtube Link, Paper, Nanohub (Hands-on) |
Potentials for Molecular Dynamics |
|
5/1 |
Youtube Link, Youtube Link |
Reinforcement Learning |
|
5/6 |
Paper, Youtube Link, Youtube Link |
Self-Driving Lab Example, Recurrent Neural Networks, Long Short Term Memory |
|
Course Description
This course is an in-depth introduction into the use of machine-learning techniques in the field of materials science.
Necessary foundations will be introduced, including how to curate and maintain data in databases, how to visualize it, and how to turn it into a descriptor compatible with machine-learning models.
Supervised, unsupervised, and reinforcement techniques to machine learning will be introduced, explained, and students will use these techniques themselves on various data sets.
Regression, classification, and deep learning models will be introduced and used by students.
The course will also explain modern applications of machine learning in the field of materials science using recent research papers.
Hands-on tutorials will be in Python.
Scope
- Statistics in Python
- Machine learning models and their application to data
- Data sources, curation, management, and visualization
- Applications from recent research literature
Course Grading
Grading
Attendance of each unit will be graded through a GradeScope assignment, filled out in class.
Class participation on CampusWire will be graded, and it is expected that you ask two questions throughout the semester and answer two questions throughout the semester.
Several of the units will involve hands-on sessions (check the "Notes/Assignments" column above) and it is required that you to submit a jupyter notebook of your activities, which will be graded.