aeon-toolkit / aeon

A toolkit for machine learning from time series
https://aeon-toolkit.org/
BSD 3-Clause "New" or "Revised" License
912 stars 98 forks source link

[ENH] Implement Sequence Weighted Alignment model (Swale) distance function #425

Open TonyBagnall opened 1 year ago

TonyBagnall commented 1 year ago

Describe the feature or idea you want to propose

Swale is a distance function used in some comparisons, for example here https://www.cs.ucr.edu/~eamonn/vldb_08_Experimental_comparison_time_series.pdf https://dl.acm.org/doi/10.1145/1247480.1247544 (there is another more recent comparison I cant find the link for right now, will post it anon)

We have not used Swale in any of our distance experiments, because it did not perform very well in the above study. However, that was 15 years ago and it would be timely to re-evaluate it through an aeon implementation.

Describe your proposed solution

A numba based function modelled along the lines of existing distances. I can probably track down a matlab implementation if someone wants to take this on. there is a link to a java implementation here http://users.ece.northwestern.edu/~hdi117/tsim.htm

Describe alternatives you've considered, if relevant

No response

Additional context

No response

TonyBagnall commented 3 months ago

journal version of comparison https://link.springer.com/article/10.1007/s10618-012-0250-5