This short course introduces the core concepts and intuitions of the three most important branches of mathematics for machine learning: linear algebra, calculus, and probability.
These notebooks are supplementary to a YouTube lecture series and presume proficiency with Python.
The notebooks can be executed in any of the following ways.
| Notebook | Link |
|---|---|
| Linear Algebra |  |
| Calculus | |
| Probability | |
Click the badges in the table above to access the exercises as Google Colab notebooks. The only requirement is a Google account. If you save your work to your own Google Drive, you can return where you left off.
If you click the badge above, you'll launch a free cloud server provided by the Binder project with the appropriate computational environment. There is no need to create an account. This environment is ephemeral, or temporary: after 10 minutes of inactivity, it will disappear. You'll be able to run the notebooks, but the only way to permanently save any work is to download the files to your machine. In order to continue from where you left off, you'd then need to re-upload the files to a new Binder instance.
This option is simple, sufficient for most purposes (the exercises are very short), and well-tested.
If you'd like to run the materials locally, the best option is to use Docker, one of the virtualization technologies on which Binder is based.
After following the installation instructions for Docker, build the container with the command
docker build -t math-for-ml .
and then start it with the command
docker run -p 8888:8888 math-for-ml
Open a browser window and navigate to
localhost:8888
and enter, as the password, the token that appears after
?token in the URLs printed to the terminal by Jupyter.
If you are unfamiliar with or unable to use Docker,
you can instead use pip to install the necessary packages.
They are located in requirements-local.txt
and can be installed with
pip install -r requirements-local.txt
Note that the requirements are very strictly versioned, to reduce bugs. It is highly recommended to use a virtual environment tool, like virtualenv or pyenv to set up a specific environment for use with these notebooks. In general, Python is best used with a virtual environment tool, so setting one up will brings large dividends for future projects!