Abstract:Graph convolution networks (GCNs) have achieved remarkable success in processing non-Euclidean data. GCNs update the feature representations of each sample by aggregating the structure information from K -order (layer) neighborhood samples. Existing GCNs variants rely heavily on the K -th layer semantic information with K -order neighborhood information aggregating. However, semantic features from different convolution layers have distinct sample attributes. The single-layer semantic feature is only a one-sided feature representation. Besides, the semantic features of traditional GCNs will be oversmoothing with multi-layer structure information aggregates. In this paper, to solve the above-mentioned problem, we propose adaptive graph convolutional collaboration networks (AGCCNs) for the semi-supervised classification task. AGCCNs can fully use the different scales of discrimination information contained in the different convolutional layers. Specifically, AGCCNs utilize the attention mechanism to learn the relevance (contribution) coefficient of the deep semantic features from different convolution layers for the task, which aims to effectively discriminant their importance. After multiple optimizations, AGCCNs can adaptively learn the robust deep semantic features via the effective semantic fusion process between multi-layer semantic information. Compared with GCNs that only utilize the K -th layer semantic features, AGCCNs make the learned deep semantic features contain richer and more robust semantic information. What is more, our proposed AGCCNs can aggregate the appropriate K -order neighborhood information for each sample, which can relieve the oversmoothing issue of traditional GCNs and better generalize shallow GCNs to more deep layers. Abundant experimental results on several popular datasets demonstrate the superiority of our proposed AGCCNs compared with traditional GCNs.

Attention-based Frequency Adaptation Graph Convolutional Network

Beyond Homophily and Homogeneity Assumption: Relation-Based Frequency Adaptive Graph Neural Networks

Beyond Low-frequency Information in Graph Convolutional Networks

AF-GCN: Completing Various Graph Tasks Efficiently Via Adaptive Quadratic Frequency Response Function in Graph Spectral Domain

Attention-based Graph Convolution Fusing Latent Structures and Multiple Features for Graph Neural Networks

AAGCN: a graph convolutional neural network with adaptive feature and topology learning

Adaptive Graph Diffusion Networks

Graph Adaptive Attention Network with Cross-Entropy

AF-GCN: Attribute-Fusing Graph Convolution Network for Recommendation

Attention-driven Graph Clustering Network.

Exploiting Neighbor Effect: Conv-Agnostic GNN Framework for Graphs with Heterophily

DAGCN: Dual Attention Graph Convolutional Networks

GRAF: Graph Attention-aware Fusion Networks

Adaptive Graph Convolutional Collaboration Networks for Semi-supervised Classification

AM-GCN: Adaptive Multi-channel Graph Convolutional Networks

Attribute-Aware Graph Convolutional Network Recommendation Method

DCGNN: Adaptive deep graph convolution for heterophily graphs

AGNN: Alternating Graph-Regularized Neural Networks to Alleviate Over-Smoothing

Node classification oriented Adaptive Multichannel Heterogeneous Graph Neural Network

Learning Discrete Adaptive Receptive Fields for Graph Convolutional Networks