University of Maryland's AOSC 650 course: Neural Networks for the Physical Sciences
In this course, we will learn the fundamentals of neural networks, from the basics of a neuron to more advanced architectures, and how these tools can be applied to the physical sciences. Important considerations for data preprocessing specific to the physical sciences will be discussed, along with how to evaluate the skill, uncertainty, and confidence of neural networks with specific relevance to the physical sciences.
The course syllabus is available here.
To use this repository for the course, fork your own copy of the repository.
Then, your fork of the repository can be downloaded to your local machine using git on your terminal:
git clone https://github.com/$YOUR_GITHUB_USERNAME$/UMDAOSC650.git
To install the necessary python environment, using the provided yaml file is strongly recommended.
Run the following command from within the UMDAOSC650 directory to install:
conda env create -f keras-tf.yml
Once installed, the python environment can be activated by running:
conda activate keras-tf-v2025
To deactivate the python environment, run:
conda deactivate