4 papers · 1 filter
Differences Between Hard and Noisy-labeled Samples: An Empirical Study
Mahsa Forouzesh, Patrick Thiran
Extracting noisy or incorrectly labeled samples from a labeled dataset with hard/difficult samples is an important yet under-explored topic. Two general and often independent lines…
Leveraging Unlabeled Data to Track Memorization
Mahsa Forouzesh, Hanie Sedghi, Patrick Thiran
Deep neural networks may easily memorize noisy labels present in real-world data, which degrades their ability to generalize. It is therefore important to track and evaluate the ro…
Disparity Between Batches as a Signal for Early Stopping
Mahsa Forouzesh, Patrick Thiran
We propose a metric for evaluating the generalization ability of deep neural networks trained with mini-batch gradient descent. Our metric, called gradient disparity, is the $\ell_…
Generalization Comparison of Deep Neural Networks via Output Sensitivity
Mahsa Forouzesh, Farnood Salehi, Patrick Thiran
Although recent works have brought some insights into the performance improvement of techniques used in state-of-the-art deep-learning models, more work is needed to understand the…