-
PH6232: Machine Learning and its physics applications
-
Course Website: https://alaha999.github.io/
-
Course Modules: https://alaha999.github.io/modules/
We need python,numpy,matplotlib,tensorflow,keras, and scikit-learn for this course.
- How to install scikit-learn: install scikit-learn
- How to install tensorflow: install_tf_documentation, pypi_pip, how-to-install-python-tensorflow-in-windows
- How to install Keras: install keras
NB: Please make sure you have these libraries working in your system. If not, make sure you come to Wednesday's tutorial session, and we will help you with the installation! But this comes with a free joke,
Use the conda distribution to streamline your workflow and get rid of frequent anxiety attacks due to package dependency errors, etc. The steps are the following,
- Install conda: conda documentation [ You can use miniconda or mamba whatever you like]
- Make a conda environment: Managing environments (say the name is
ml_course) - Activate the conda environment:
conda activate ml_course - Then inside this environment, install the packages:
conda install <package-name>
Note: Each time you want to work, you activate the environment and proceed. Installing packages is only for once.
| Week | Materials |
|---|---|
| Week 1 | - Python basics, matplotlib, numpy, pandas dataframe, etc. - Lectures on neural networks (overview and maths) |
| Week 2 | - Regression using DNN in the context of projectile motion: Lectures and hands-on session |
| Week 3 | - Classification using DNN in the context of projectile motion: Lectures and hands-on session |
| Week 4 | - Assignments: WZ vs ZZ classification (LHC problem) and Gravitation Wave related classification (LIGO) |
| Week 5 | - Convolutional neural network (CNN): Lectures and hands-on session (Low Pt vs High Pt track) |
| Week 6 | - Unsupervised ML and Dimension reduction techniques |
| Week 7 | - Generative Adversarial Networks (GAN) |
| Week 8 | - Guest Lectures: Anomaly detection in the finance sector and Quantum ML |
Author:
Contact Arnab Laha ([email protected]) for any details. Or visit office A-365, Main Building Second Floor.