[NewModel] ROCKET Classification

[JanFidor]

https://arxiv.org/abs/1910.13051

This approach feels very unique compared to the majority of forecasting models which are either using Transformers or DNNs while also being much faster.

It's for TS classification instead of forecasting, so I'm not certain if it would be a good fit for darts. Actually after some more research it seems someone already tried to adapt it for regression https://link.springer.com/article/10.1007/s10618-021-00745-9

Related to https://github.com/unit8co/darts/issues/1473

[solalatus]

There is also an "updated" version of it called MINIROCKET which offers considerable speed advantages. Here is the reference implementation.

[hrzn]

Thanks for the suggestion, adding to backlog.