STAR-GCN：推荐系统中的堆叠重建图卷积网络

需积分: 0 34 浏览量更新于2024-08-05 收藏 337KB PDF 举报

"2019年提出的STAR-GCN（Stacked and Reconstructed Graph Convolutional Network）是一种用于推荐系统的深度学习模型，旨在提高在冷启动场景下的性能。该模型由Jiani Zhang、Xingjian Shi、Shenglin Zhao和Irwin King共同研发，分别来自香港中文大学、香港科技大学和腾讯优图实验室。STAR-GCN通过堆叠的GCN编码器-解码器结构和中间监督来提升预测性能，同时解决了传统图卷积网络在处理冷启动问题时的局限性。在传统的图卷积网络中，节点输入通常采用一热编码，这可能导致模型空间复杂度增加。而STAR-GCN则采用低维的用户和物品潜在因子作为输入，有效地限制了模型的复杂度。这种设计允许网络学习更丰富的节点表示，从而更好地捕捉用户和物品之间的关系。 STAR-GCN的核心创新在于其重建机制。通过对输入节点嵌入进行掩模操作并重建，模型能够为新出现的节点生成嵌入，这直接解决了推荐系统中的冷启动问题。在没有历史交互数据的情况下，新用户的推荐和新物品的引入变得可能，提高了系统的适应性和泛化能力。此外，STAR-GCN的堆叠结构增强了模型的表达能力。每一层的GCN编码器负责从上一层的节点特征中学习信息，而解码器则将这些信息转换回原始空间，通过中间监督来指导学习过程。这样的设计有助于捕获更深层次的依赖关系，并在多层传播中逐渐细化节点的表示。在推荐系统中，STAR-GCN的性能提升主要体现在两个方面：一是通过低维度的潜在因子学习，减少了过拟合的风险，提升了模型的泛化性能；二是通过重建机制，它能够在没有足够历史数据的情况下为新节点提供有效的嵌入，从而改善了冷启动问题的处理。因此，STAR-GCN是推荐系统领域中一个重要的进展，尤其适用于需要处理大量新用户和新物品的实时推荐场景。"

STAR-GCN: Stacked and Reconstructed Graph Convolutional Networks for

Recommender Systems

Jiani Zhang

, Xingjian Shi

, Shenglin Zhao

and Irwin King

The Chinese University of Hong Kong, Hong Kong, China

Hong Kong University of Science and Technology, Hong Kong, China

Youtu Lab, Tencent, Shenzhen, China

{jnzhang, king}@cse.cuhk.edu.hk, xshiab@connect.ust.hk, zsl.zju@gmail.com

Abstract

We propose a new

STA

cked and

econstructed

raph

onvolutional

etworks (STAR-GCN) ar-

chitecture to learn node representations for boost-

ing the performance in recommender systems, es-

pecially in the cold start scenario. STAR-GCN em-

ploys a stack of GCN encoder-decoders combined

with intermediate supervision to improve the ﬁnal

prediction performance. Unlike the graph convo-

lutional matrix completion model with one-hot en-

coding node inputs, our STAR-GCN learns low-

dimensional user and item latent factors as the input

to restrain the model space complexity. Moreover,

our STAR-GCN can produce node embeddings for

new nodes by reconstructing masked input node em-

beddings, which essentially tackles the cold start

problem. Furthermore, we discover a label leak-

age issue when training GCN-based models for

link prediction tasks and propose a training strat-

egy to avoid the issue. Empirical results on multi-

ple rating prediction benchmarks demonstrate our

model achieves state-of-the-art performance in four

out of ﬁve real-world datasets and signiﬁcant im-

provements in predicting ratings in the cold start

scenario. The code implementation is available in

https://github.com/jennyzhang0215/STAR-GCN.

1 Introduction

Recommender systems, which analyze users’ preference pat-

terns to suggest potential targets, are indispensable in content

providers, electronic retailers, web search engines, etc. The

key mathematical problem underlying recommender systems

is matrix completion

[

Cand

es and Recht, 2009

]

. Assume there

are

users and

items, the recommendation algorithm aims

to ﬁll in the missing entries in the

N × M

rating matrix given

the existing entries.

The classical way to solve this problem is via Matrix Factor-

ization (MF)

[

Koren et al., 2009

]

, in which the rating scores are

generated by functions over the latent factors or embeddings

of users and items. Recent advancements in deep learning,

especially Graph Convolutional Networks (GCN)

[

Defferrard

et al., 2016; Bronstein et al., 2017; Kipf and Welling, 2017;

Hamilton et al., 2017

]

, have brought new ideas for tackling

this essential artiﬁcial intelligence problem. GCN general-

izes the deﬁnition of convolution from the regular grid to

irregular grid, like graph structures. The GCN framework gen-

erates node representations by a localized parameter-sharing

operator, known as graph aggregator

[

Hamilton et al., 2017;

Zhang et al., 2018

]

. A graph aggregator calculates a node’s

representation by transforming and aggregating the features of

its local neighborhoods. By stacking multiple graph aggrega-

tors and nonlinear functions, we build a deep neural network

that can extract features across far reaches of a graph. Because

the local neighborhood set can be viewed as the receptive ﬁeld

of a convolution kernel, this kind of neighborhood aggrega-

tion methods is named as graph convolution, which also have

connections to spectral graph theory

[

Kipf and Welling, 2017

]

Monti et al.

[

2017

]

proposed the ﬁrst GCN-based method

for recommender systems. In their approach, GCN was used

to aggregate information from two auxiliary user-user and

item-item graphs. The latent factors of users and items were

updated after each aggregation step, and a combined objective

function of GCN and MF was used to train the model. After

that, Berg et al.

[

2017

]

proposed the Graph Convolutional

Matrix Completion (GC-MC) model. GC-MC directly charac-

terized the relationship between users and items as a bipartite

interaction graph. Two multi-link graph convolution layers

were used to aggregate user features and item features. The rat-

ings were estimated by predicting the edge labels. Thanks to

the power of GCN in learning high-quality user and item repre-

sentations, GC-MC has achieved state-of-the-art performance

in several public recommendation benchmarks.

While being powerful, the GC-MC model has two signif-

icant limitations. To distinguish each node, the model uses

one-hot vectors as node input. This makes the input dimen-

sionality proportional to the total number of nodes and thus

is not scalable to large graphs. Moreover, the model is unable

to predict the ratings for new users or items that are not seen

in the training phase because we cannot represent unknown

nodes as one-hot vectors. The task of predicting ratings for

new users or items is also known as the cold start problem.

In this paper, we propose a new architecture,

STA

cked and

econstructed

raph

onvolutional

etworks (STAR-GCN),

to solve these problems. Unlike GC-MC, STAR-GCN directly

learns low-dimensional user and item embeddings as the in-

put to the network in an end-to-end fashion. To improve the

learned embeddings and also generalize the model to predict

arXiv:1905.13129v1 [cs.IR] 27 May 2019

下载后可阅读完整内容，剩余6页未读，立即下载

网络小精灵

粉丝: 36
资源: 334

STAR-GCN：推荐系统中的堆叠重建图卷积网络

2019-ICLR-Confidence-based Graph Convolutional Networks for Semi

semi -supervised classification with graph convolutional networks学习必记

[GCN] 代码解析 of GitHub：Semi-supervised classification with graph convolutional networks

BERT-BiLSTM-GCN模型中GCN的缺点是什么

R-GCN算法与GCN算法的比较

T-GCN代码GCN几层？

简述《Multi-Label Image Recognition with Graph Convolutional Networks》的内容

CTR-GCN和TE-GCN在各数据集的表现

SEMI-SUPERVISED CLASSIFICATION WITH GRAPH CONVOLUTIONAL NETWORKS 代码

Semi-Supervised Classification with Graph Convolutional Networks

最新资源