As a reviewer, chair and general academic, I’m beginning to notice a pattern where a paper will get rejected from NeurIPS, ICLR, ICML only to end up in PNAS later on. I won’t name specific papers, but it has happened at least on a few occasions directly in my own experience. Furthermore, the are not any substantial modifications to the work either. Has anyone else noticed this pattern as well?

PNAS is a reputable journal, but they do not have the computational/technical reviewer pools that are aware of the work going on in the AI/ML community. So I suspect that some are taking advantage of this.