Towards Better Dynamic Graph Learning: New Architecture and Unified Library

Authors: Le Yu, Leilei Sun, Bowen Du, Weifeng Lv

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental By performing exhaustive experiments on thirteen datasets for dynamic link prediction and dynamic node classification tasks, we find that Dy GFormer achieves state-of-the-art performance on most of the datasets, demonstrating its effectiveness in capturing nodes correlations and long-term temporal dependencies.
Researcher Affiliation Academia Le Yu, Leilei Sun , Bowen Du, Weifeng Lv State Key Laboratory of Software Development Environment School of Computer Science and Engineering Beihang University {yule,leileisun,dubowen,lwf}@buaa.edu.cn
Pseudocode No The paper describes its methods using prose and mathematical equations, but does not include any explicit pseudocode or algorithm blocks.
Open Source Code Yes All the used resources are publicly available at https://github.com/yule-BUAA/Dy GLib.
Open Datasets Yes We experiment with thirteen datasets (Wikipedia, Reddit, MOOC, Last FM, Enron, Social Evo., UCI, Flights, Can. Parl., US Legis., UN Trade, UN Vote, and Contact), which are collected by [44] and cover diverse domains.
Dataset Splits Yes For both tasks, we chronologically split each dataset with the ratio of 70%/15%/15% for training/validation/testing.
Hardware Specification Yes Experiments are conducted on an Ubuntu machine equipped with one Intel(R) Xeon(R) Gold 6130 CPU @ 2.10GHz with 16 physical cores. The GPU device is NVIDIA Tesla T4 with 15 GB memory.
Software Dependencies No The paper mentions implementation using PyTorch ('which are all implemented by Py Torch'), but does not specify its version or other software dependencies with version numbers.
Experiment Setup Yes We set the learning rate and batch size to 0.0001 and 200 for all the methods on all the datasets.