On the reliability of crowd-sourced investigations of scientific misconduct in biomedical imagery

Image credit: H.Farid

Abstract

We describe a technique for measuring image similarity in biomedical imagery (in particular, western blots. We also describe a large-scale comparison of image similarity across authentic images and synthetically duplicated and in-the-wild duplicated images we provide a systematic, quantitative approach for assessing claims of scientific misconduct in digital imagery

Justin D. Norman
Justin D. Norman
PhD Student | ML/AI Consultant