Collaboration with Tempus

[asenzz]

Description

Hi guys, I'm the author of Tempus, a new type of SVM implementation that scales accuracy by nesting kernel matrices. A kernel function can be any model eg. LightGBM, TFT or a conventional kernel eg. global alignment or path. Tempus achieves significantly better accuracy than LightGBM in the tests so far, even with a single layer kernel function.

Use case

Would you be interested in some sort of collaboration, especially in the feature engineering department where Tempus is lacking as I think?