Computer Vision News 44 Datasets through the LookingGlass is a webinar series dedicated to exploring the datarelated aspects of machine learning methods. “Our goal is to foster a community of researchers passionate about comprehending the profound impact of the data we employ on both algorithms and society, going beyond mere performance optimization. Our inspiration draws from a diverse rangeof draws from a diverse range of subjects, encompassing data curation for dataset creation, metadata, shortcuts, fairness, ethics, and the philosophical dimensions of AI.” The webinar is part of “Making MetaDataCount” project and is organized by Veronika Cheplygina (left in the photo) and Amelia Jiménez-Sánchez (right) at IT University of Copenhagen. The goals of the project involve the investigation of different types of shortcuts (based on demographics or image artifacts) that might occur and how these affect the performance and fairness of the algorithms, as well as investigate metadata-aware methods to avoid learning such biases or shortcuts. Datasets through the L king-Glass
RkJQdWJsaXNoZXIy NTc3NzU=