ConvolutionalNeuralNetworks(CNNs)achieveimpressiveperformanceinawidevarietyoffields.Theirsuccess benefitedfromamassiveboostwhenverydeepCNNmodels wereabletobereliablytrained.Despitetheirmerits,CNNs failtoproperlyaddressproblemswithnon-Euclideandata. Toovercome