All posts by Jacek Szczerbiński

Trust no one, not even your training data! Machine learning from noisy data

Label noise is ever-present in machine learning practice. Allegro datasets are no exception. We compared 7 methods for training classifiers robust to label noise. All of them improved the model’s performance on noisy datasets. Some of the methods decreased the model’s performance in the absence of label noise.

Jacek Szczerbiński

Jacek Szczerbiński obtained his PhD in Chemistry from ETH Zurich. He then fell in love with ML and became a Research Engineer at Allegro. Currently he is studying robustness of text classifiers against mislabeled training data. His superpower is explaining ML to non-technical people.