Do spectral cues matter in contrast-based graph self-supervised learning?

Open in new window