distributed-learning-contributivity icon indicating copy to clipboard operation
distributed-learning-contributivity copied to clipboard

Published 20 hours ago • verified publisher icon LabeliaLabs

Gaussian noise dataset for corrupted partner

Open arthurPignet opened this issue 5 years ago • 1 comments
trafficstars

Today, to simulate a corrupted partner we can set all the labels of its dataset to 1 or shuffle its labels. It could be really interesting if the whole dataset of a corrupted partner can be randomly generated (Features or images, and label)

arthurPignet avatar Sep 21 '20 09:09 arthurPignet

Interesting indeed. I had not thought about noise on the features... it raise a lot of questions.

Thomas-Galtier avatar Sep 23 '20 10:09 Thomas-Galtier