Transitive Invariance for Self-supervised Visual Representation Learning | Read Paper on Bytez