English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The performance limit of Graph Convolutional Networks (GCNs) and the factthat we cannot stack more of them to increase the performance, which we usuallydo for other deep learning paradigms, are pervasively thought to be caused bythe limitations of the GCN layers, including insufficient expressive power,etc. However, if so, for a fixed architecture, it would be unlikely to lowerthe training difficulty and to improve performance by changing only thetraining procedure, which we show in this paper not only possible but possiblein several ways. This paper first identify the training difficulty of GCNs fromthe perspective of graph signal energy loss. More specifically, we find thatthe loss of energy in the backward pass during training nullifies the learningof the layers closer to the input. Then, we propose several methodologies tomitigate the training problem by slightly modifying the GCN operator, from theenergy perspective. After empirical validation, we confirm that these changesof operator lead to significant decrease in the training difficulties andnotable performance boost, without changing the composition of parameters. Withthese, we conclude that the root cause of the problem is more likely thetraining difficulty than the others.

Training Matters: Unlocking Potentials of Deeper Graph Convolutional Neural Networks