Robust Training with Ensemble Consensus

Jisoo Lee, Sae-Young Chung

Since deep neural networks are over-parameterized, they can memorize noisy examples. We address such a memorization issue in the presence of label noise. From the fact that deep ne…