Partially trained graph convolutional networks resist oversmoothing
Sign Up to like & getrecommendations! Published in 2024 at "Machine Learning"
DOI: 10.1007/s10994-025-06865-3
Abstract: In this work we investigate an observation made by Kipf and Welling (5th International Conference on Learning Representations, 2017), who suggested that untrained Graph Convolutional Networks (GCNs) can generate meaningful node embeddings. In particular, we… read more here.
Keywords: graph convolutional; networks resist; partially trained; resist oversmoothing ... See more keywords