[论文解读] Revisiting Graph based Collaborative Filtering: A Linear Residual Graph Convolutional Network Approach

Lei Chen, Le Wu+3 more|arXiv (Cornell University)|Jan 28, 2020

Recommender Systems and Techniques参考文献 34被引用 13

一句话总结

该论文提出LR-GCCF，一种用于协同过滤的线性残差图卷积网络，通过移除非线性激活函数并引入残差学习，以缓解深层GCN中的过平滑问题。该模型在大规模稀疏用户-物品图上实现了更快的训练速度和更好的可扩展性，且在Amazon Books和Gowalla数据集上优于现有的基于GCN的方法。

ABSTRACT

Graph Convolutional Networks (GCNs) are state-of-the-art graph based representation learning models by iteratively stacking multiple layers of convolution aggregation operations and non-linear activation operations. Recently, in Collaborative Filtering (CF) based Recommender Systems (RS), by treating the user-item interaction behavior as a bipartite graph, some researchers model higher-layer collaborative signals with GCNs. These GCN based recommender models show superior performance compared to traditional works. However, these models suffer from training difficulty with non-linear activations for large user-item graphs. Besides, most GCN based models could not model deeper layers due to the over smoothing effect with the graph convolution operation. In this paper, we revisit GCN based CF models from two aspects. First, we empirically show that removing non-linearities would enhance recommendation performance, which is consistent with the theories in simple graph convolutional networks. Second, we propose a residual network structure that is specifically designed for CF with user-item interaction modeling, which alleviates the over smoothing problem in graph convolution aggregation operation with sparse user-item interaction data. The proposed model is a linear model and it is easy to train, scale to large datasets, and yield better efficiency and effectiveness on two real datasets. We publish the source code at https://github.com/newlei/LRGCCF.

研究动机与目标

• 解决基于深层GCN的协同过滤模型中的训练困难和过平滑问题。
• 探究非线性激活函数在基于GCN的推荐系统中是否必要。
• 设计一种残差学习机制，在建模高阶协同信号的同时保留用户独特性。
• 为大规模推荐系统开发一种可扩展、高效且有效的线性GCN模型。

提出的方法

• 提出一种线性图卷积网络（L-GCCF），用线性嵌入传播替代非线性激活函数，以降低复杂度并提升训练效率。
• 引入一种残差偏好学习模块，显式在每一层建模用户-物品偏好差异，以对抗过平滑。
• 采用受ResNet启发的残差连接结构，其中每一层学习当前嵌入相对于前一层表示的残差。
• 在用户-物品二分图中对K跳邻域进行递归邻居聚合，以捕捉高阶协同信号。
• 通过无非线性性的线性变换实现嵌入传播，简化优化过程。
• 将线性传播与残差学习相结合，联合保留用户独特性并建模深层协同模式。

实验结果

研究问题

• RQ1 在基于GCN的协同过滤中移除非线性激活是否能提升推荐性能和训练稳定性？
• RQ2 残差学习机制是否能有效缓解稀疏用户-物品交互图中深层GCN的过平滑问题？
• RQ3 当结合线性传播与残差学习时，图卷积层的深度如何影响性能？
• RQ4 在线性GCN模型中，是否能在大规模推荐系统中实现比非线性GCN模型更优的效率和可扩展性？

主要发现

• 在基于GCN的CF模型中移除非线性激活可提升推荐性能并加快训练速度，与简化GCN中的理论发现一致。
• 所提出的LR-GCCF模型在Amazon Books和Gowalla数据集上均达到最先进性能，优于NGCF、PinSage及其他基于GCN的基线模型。
• 在Amazon Books上，LR-GCCF在K=4时达到HR@20为0.0341，NDCG@20为0.0258，优于所有测试模型。
• 在Gowalla上，最优性能在K=3时达到，HR@20为0.1518，NDCG@20为0.1259。
• 在Amazon Books上，LR-GCCF在K=4时每轮训练约需20秒，显著快于PinSage（38秒）和NGCF（38秒）在K=2时的训练时间。
• LR-GCCF中用户嵌入的余弦相似度方差高于非残差变体（L-GCCF），实证验证了残差学习可减少过平滑并保留用户独特性。