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.
In July we attended the scientific conference SIGIR 2016 held in Pisa, Italy. SIGIR is an annual conference related to new advances in Information Retrieval. The shortcut is for Special Interest Group on Information Retrieval. This is an annual conference with the highest ranking at Microsoft Academic Research in Information Retrieval field field and the 16th in the ranking in the whole Computer Science field.
Senior Research Engineer in the Machine Learning Research team at Allegro, where she works on applying and advancing NLP methods in the e-commerce domain. Obtained her PhD from the University of Warsaw, where she focused on machine learning methods for histopathology.
