This webpage supports our paper "Domain Agnostic Online Semantic Segmentation"

The code is here.

The datasets are here.

The key to explain the datasets is here.

Supporting materials are here, including a comparasion to AutoPlait on their data, and an example of FLOSS on multidimensional data.

We had to adapt the AutoPlait code slightly, here is our modified version, and the classic version

The ideas are built on top the Matrix Profile.

Here is a short video that shows FLOSS running in real time, on a toy dataset. Note that the processing rate is slowed down to help us see the algorithm in action. FLOSS can process streaming data at least an order of magniture faster.