Handling Unbalanced Data

[richie]

Hi!

First off, thanks for the continuing development of MEPX!

In continuation of my previous post on "Special points in a timeseries," I'm pondering a broader aspect, concerning unbalanced data. While previously I focused on pinpointing specific points in a timeseries, I'm beginning to wonder in general about MEPX's capabilities in dealing with data imbalance.

Is MEPX already equipped to adjust its learning process to assign more weight to rare values during training? I haven't come across such functionality in my exploration of the program. Could implementing techniques like class weighting potentially address this issue?

Furthermore, I'm curious about the choice of loss functions. Are there alternative options that could prioritize the correct classification of rare values, considering the challenges posed by unbalanced data? Would it make sense to explore specialized loss functions for handling data imbalance within MEPX?

Keep up the good work!

Kind regards,

Richie

[Mihai Oltean]

Hi,

Currently there is no mechanism to handle very unbalanced data.

I'll think more about it.

best regards,

--
You received this message because you are subscribed to the Google Groups "Multi Expression Programming" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mepx+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mepx/37ef99e5-f635-4498-bdad-a3d5e5fbd5c2n%40googlegroups.com.