Bag of Tricks for Node Classification with Graph Neural Networks

Wang, Yangkun; Jin, Jiarui; Zhang, Weinan; Yu, Yong; Zhang, Zheng; Wipf, David

doi:10.48550/arxiv.2103.13355

Cited by 10 publications

(18 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this paper, we combine the advantages from both search-based and time-aware models to efficiently retrieve relevant items and mine sequential patterns in an end-to-end way. Our paper is also related to the label trick proposed in [38,42] based graph structure. Instead, our work focuses on the label usage in the sequence cases, which, notably, is also different from the masking technique in existing sequential models such as BERT [7] performing on the feature dimension instead of the label dimension.…”

Section: Preliminaries 21 Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Learn over Past, Evolve for Future: Search-based Time-aware Recommendation with Sequential Behavior Data

Jin¹,

Chen²,

Zhang³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

The personalized recommendation is an essential part of modern ecommerce, where user's demands are not only conditioned by their profile but also by their recent browsing behaviors as well as periodical purchases made some time ago. In this paper, we propose a novel framework named Search-based Time-Aware Recommendation (STARec), which captures the evolving demands of users over time through a unified search-based time-aware model. More concretely, we first design a search-based module to retrieve a user's relevant historical behaviors, which are then mixed up with her recent records to be fed into a time-aware sequential network for capturing her time-sensitive demands. Besides retrieving relevant information from her personal history, we also propose to search and retrieve similar user's records as an additional reference. All these sequential records are further fused to make the final recommendation. Beyond this framework, we also develop a novel label trick that uses the previous labels (i.e., user's feedbacks) as the input to better capture the user's browsing pattern. We conduct extensive experiments on three real-world commercial datasets on click-through-rate prediction tasks against state-of-the-art methods. Experimental results demonstrate the superiority and efficiency of our proposed framework and techniques. Furthermore, results of online experiments on a daily item recommendation platform of Company X show that STARec gains average performance improvement of around 6% and 1.5% in its two main item recommendation scenarios on CTR metric respectively.

show abstract

Section: Preliminaries 21 Related Workmentioning

confidence: 99%

“…The principal way to use the user historical feedbacks is to treat this feedbacks as the label to supervise the model. However, as discussed in [38,42], combining the information from both label and feature as the input to train the model can significantly improve its performance.…”

Section: Introductionmentioning

confidence: 99%

Learn over Past, Evolve for Future: Search-based Time-aware Recommendation with Sequential Behavior Data

Jin¹,

Chen²,

Zhang³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…GAT-FLAG GAT with FLAG [15] enhancement. GAT+BoT GAT with bag of tricks [32]. AGDN+BoT AGDN with bag of tricks.…”

Section: Gnn Model Descriptionmentioning

confidence: 99%

“…But as the baseline performance of reddit dataset is quite high, not surprisingly, the overall improvement of NGNN is not significant. We further analysis the performance of NGNN combined with bag of tricks [32] on ogbn-arxiv and ogbn-proteins in Table 4. It can Table 5: Performance of NGNN on ogbl-collab, ogbl-ppa and ogbl-ppi.…”

Section: Node Classificationmentioning

confidence: 99%

“…It improves the test accuracy by 1.6% on the ogbn-products datasets for GraphSage. Furthermore, NGNN with AGDN+BoT+self-KD+C&S [13] achieves the forth place on the ogbn-arxiv leaderboard 1 and NGNN with GAT+BoT [32] achieves second place on the ogbn-proteins leaderboard with many fewer model parameters. • NGNN improves the performance of SEAL, GCN and Graph-Sage and their variants on link prediction datasets including ogbl-collab, ogbl-ppa and ogbl-ppi.…”

Section: Introductionmentioning

confidence: 98%

See 1 more Smart Citation

Network In Graph Neural Network

Song¹,

Ma²,

Li³

et al. 2021

Preprint

View full text Add to dashboard Cite

Graph Neural Networks (GNNs) have shown success in learning from graph structured data containing node/edge feature information, with application to social networks, recommendation, fraud detection and knowledge graph reasoning. In this regard, various strategies have been proposed in the past to improve the expressiveness of GNNs. For example, one straightforward option is to simply increase the parameter size by either expanding the hidden dimension or increasing the number of GNN layers. However, wider hidden layers can easily lead to overfitting, and incrementally adding more GNN layers can potentially result in over-smoothing. In this paper, we present a model-agnostic methodology, namely Network In Graph Neural Network (NGNN ), that allows arbitrary GNN models to increase their model capacity by making the model deeper. However, instead of adding or widening GNN layers, NGNN deepens a GNN model by inserting non-linear feedforward neural network layer(s) within each GNN layer. Although, some works mentioned that adding MLPs within GNN layers could benefit the performance, they did not systematically analyze the reason for the improvement, nor evaluate with numerous GNNs on large-scale graph datasets. In this paper, we demonstrate that NGNN can keep the model stable against either node feature or graph structure perturbations through an analysis of it as applied to a GraphSage base GNN on ogbn-products data. Furthermore, we take a wideranging evaluation of NGNN on both node classification and link prediction tasks and show that NGNN works reliably across diverse GNN architectures. For instance, it improves the test accuracy of GraphSage on the ogbn-products by 1.6% and improves the hits@100 score of SEAL on ogbl-ppa by 7.08% and the hits@20 score of GraphSage+Edge-Attr on ogbl-ppi by 6.22%. And at the time of this submission, it achieved two first places on the OGB link prediction leaderboard.

show abstract

Open Biomedical Network Benchmark A Python Toolkit for Benchmarking Datasets with Biomedical Networks

Liu

Krishnan

2023

Preprint

View full text Add to dashboard Cite

Recent breakthroughs in graph representation learning (GRL) methods have vastly advanced many real-world applications where graph structures naturally arise, such as drug discoveries, molecular property predictions, and social recommendation systems. The rapid developments of GRL with applications in specific scientific domains are largely enabled by graph learning libraries such as PyTorch Geometric, which provide infrastructures that allow domain scientists to share domain-specific benchmarking datasets, and GRL researchers to adapt and design specialized GRL methods for these particular tasks. Meanwhile, over the past two decades, network biology has demonstrated superior value in harvesting biological insights from network data, such as identifying genes' associated functions, diseases, and traits. Nevertheless, the GRL community faces a significant barrier in working with biological networks because of the tedious and specialized (pre-)processing steps required to set up machine learning (ML)-ready benchmarking datasets. Here, we present nleval, a Python package containing reusable modules that enable researchers to effortlessly set up PyG-compatible ML-ready datasets using data downloaded from public databases. We expect nleval to help network biologists set up custom benchmarking datasets for answering specific biological questions of their interests and help GRL researchers adapt these datasets for designing new specialized GRL architectures.

show abstract

Bag of Tricks for Node Classification with Graph Neural Networks

Cited by 10 publications

References 18 publications

Learn over Past, Evolve for Future: Search-based Time-aware Recommendation with Sequential Behavior Data

Learn over Past, Evolve for Future: Search-based Time-aware Recommendation with Sequential Behavior Data

Network In Graph Neural Network

Open Biomedical Network Benchmark A Python Toolkit for Benchmarking Datasets with Biomedical Networks

Contact Info

Product

Resources

About