News for hackers. Stuff that matters.
Re:Training computer vision models on random noise instead o... by S_A_P on Thursday December 09, 2021 @11:25PM (#29506376)
To me it seems like feeding garbage to train the model would expose any sort of ‘bias’ the model has no?
Re:Training computer vision models on random noise instead o... by framecowbird on Friday December 10, 2021 @01:01AM (#29506883)
I'm no ML expert, but my understanding is that the bias is usually in the training data, not the model. The idea here is that you can generate bias-free "garbage" and therefore reduce the bias of the trained model
Re:Training computer vision models on random noise instead o... by SailingSperm on Friday December 10, 2021 @01:58AM (#29507189)
Yeah, very interesting... Like you could do a preliminary training that uses really good quality labeled data. Then almost infinitely continue training on a negative only dataset of random noise, which would theoretically hammer out any false positives that can be generated by the model.