Can contrastive learning avoid shortcut solutions?
Joshua Robinson, Sun Li, Ke Yu, Kayhan N. Batmanghelich, Stefanie Jegelka, Suvrit Sra · PubMed · 2021
(IFM), a method for altering positive and negative samples in order to guide contrastive models towards capturing a wider variety of predictive features. Empirically, we observe that IFM reduces feature suppression, and as a result improves performance on vision and medical imaging tasks. The code is available at: https://github.com/joshr17/IFM.