Subset Extended Kalman Filter

Updating Neural Network models as changes in the underlying system occur:

• slowly and monotonically (drift)
• discretely (transfer learning)

Installation

The only required dependencies aside from the standard library are numpy>=1.8.0 (for .npz file saving) and torch>=2.0.0 (for torch.func functions)

The sekf package can be installed using pip:

• pip install sekf to install from pypi
• pip install git+https://github.com/joshuaeh/Subset-Extended-Kalman-Filter.git to install from the github repository
• clone the repository using git clone https://github.com/joshuaeh/Subset-Extended-Kalman-Filter.git, then install locally using pip install -e <path_to_repo>
• If you are getting a ModuleNotFoundError, check that you are using the correct python environment. I've also found it helpful to declare the package name when installing locally using uv: uv pip install -e "sekf @ .".

TODO

• documentation