Decouple Graph Neural Networks: Train Multiple Simple GNNs Simultaneously Instead of One
This addresses the problem of time-consuming GNN training for researchers and practitioners, but it is incremental as it builds on existing GNN methods with a novel training framework.
The paper tackles the inefficiency of training Graph Neural Networks (GNNs) due to exponential node dependency growth by decoupling a multi-layer GNN into multiple simple modules for more efficient training, achieving high efficiency with reasonable performance as shown in theoretical and experimental results.
Graph neural networks (GNN) suffer from severe inefficiency. It is mainly caused by the exponential growth of node dependency with the increase of layers. It extremely limits the application of stochastic optimization algorithms so that the training of GNN is usually time-consuming. To address this problem, we propose to decouple a multi-layer GNN as multiple simple modules for more efficient training, which is comprised of classical forward training (FT)and designed backward training (BT). Under the proposed framework, each module can be trained efficiently in FT by stochastic algorithms without distortion of graph information owing to its simplicity. To avoid the only unidirectional information delivery of FT and sufficiently train shallow modules with the deeper ones, we develop a backward training mechanism that makes the former modules perceive the latter modules. The backward training introduces the reversed information delivery into the decoupled modules as well as the forward information delivery. To investigate how the decoupling and greedy training affect the representational capacity, we theoretically prove that the error produced by linear modules will not accumulate on unsupervised tasks in most cases. The theoretical and experimental results show that the proposed framework is highly efficient with reasonable performance.