Learning Concept Credible Models for Mitigating Shortcuts | Read Paper on Bytez