research on mitigating over-squashing of graph neural networks (gnns) has been presented. the work involves channel capacity constrained estimation as a potential solution. the research appears to be part of a technical track at an event related to artificial intelligence.