Size ratio of the forget and retain sets

[Mohamed Suliman]

Hello,

I would like to ask about the size of the forget dataset i.e the amount of data requested to be unlearnt in relation to the original, full training dataset. Is it set as a constant or can it vary, and will it be released to the pariticipants e.g 90-10, 80-20, etc?

Thanks.

[Eleni Triantafillou]

Hi,

Thanks for your question!

As an update, our competition is now open on Kaggle. See our most recent message for more info.

We have also released a document describing our evaluation procedure where we share details about the forget set size and distribution. 

Hope this helps. Please let us know if you have any other questions.

Thanks,

Eleni

--
You received this message because you are subscribed to the Google Groups "unlearning-challenge" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unlearning-chall...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/unlearning-challenge/26b31aba-bdf2-4372-86cc-3e9b613758d3n%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.