“…Generally, graph distillation draws inspiration from data distillation techniques (Wang et al 2018;Bohdal, Yang, and Hospedales 2020;Nguyen, Chen, and Lee 2021) and aims to ensure consistency between raw and synthetic datasets by constraining the soft labels across both sets. Recently, some trajectory matching algorithms show great prominence in image (Kim et al 2022;Lee et al 2022;Jiang et al 2022) and graph realms (Jin et al 2022b,a). Concretely, these frameworks adopt parameter or gradient matching scheme w.r.t the condensed set and raw data during the training process.…”