Abstract: This research proposes an approach that uses nonrepeating groups as the basis for decomposing relations. A repeating group refers to a column or set of columns that have duplicated data ...

Abstract: Data in the visual world often present long-tailed distributions. However, learning high-quality representations and classifiers for imbalanced data is still challenging for data-driven deep ...