You Can Have Better Graph Neural Networks by Not Training Weights at All: Finding Untrained GNNs Tickets

Onderzoeksoutput: Bijdrage aan tijdschriftTijdschriftartikelAcademic

42 Downloads (Pure)

Samenvatting

Recent works have impressively demonstrated that there exists a subnetwork in randomly initialized convolutional neural networks (CNNs) that can match the performance of the fully trained dense networks at initialization, without any optimization of the weights of the network (i.e., untrained networks). However, the presence of such untrained subnetworks in graph neural networks (GNNs) still remains mysterious. In this paper we carry out the first-of-its-kind exploration of discovering matching untrained GNNs. With sparsity as the core tool, we can find \textit{untrained sparse subnetworks} at the initialization, that can match the performance of \textit{fully trained dense} GNNs. Besides this already encouraging finding of comparable performance, we show that the found untrained subnetworks can substantially mitigate the GNN over-smoothing problem, hence becoming a powerful tool to enable deeper GNNs without bells and whistles. We also observe that such sparse untrained subnetworks have appealing performance in out-of-distribution detection and robustness of input perturbations. We evaluate our method across widely-used GNN architectures on various popular datasets including the Open Graph Benchmark (OGB).
Originele taal-2Engels
Artikelnummer2211.15335
TijdschriftarXiv
Volume2022
DOI's
StatusGepubliceerd - 28 nov. 2022

Bibliografische nota

Accepted by the LoG conference 2022 as a spotlight

Trefwoorden

  • cs.LG

Vingerafdruk

Duik in de onderzoeksthema's van 'You Can Have Better Graph Neural Networks by Not Training Weights at All: Finding Untrained GNNs Tickets'. Samen vormen ze een unieke vingerafdruk.
  • Best Paper Award

    Janssen, W. (Ontvanger), 2014

    Prijs: AndersWerk, activiteit of publicatie gerelateerde prijzen (lifetime, best paper, poster etc.)Wetenschappelijk

Citeer dit