GraphVite: A High-Performance CPU-GPU Hybrid System for Node Embedding

2 Mar 2019 Zhaocheng Zhu Shizhen Xu Meng Qu Jian Tang

Learning continuous representations of nodes is attracting growing interest in both academia and industry recently, due to their simplicity and effectiveness in a variety of applications. Most of existing node embedding algorithms and systems are capable of processing networks with hundreds of thousands or a few millions of nodes... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Link Prediction FB15k SimplE MR 74 # 7
MRR 0.779 # 12
Hits@10 0.876 # 14
Hits@3 0.818 # 10
Hits@1 0.721 # 8
training time (s) 2105 # 1
Link Prediction FB15k-237 RotatE MRR 0.314 # 30
Hits@10 0.511 # 29
Hits@3 0.347 # 22
Hits@1 0.217 # 26
MR 176 # 7
training time (s) 857 # 1
Link Prediction WN18 SimplE MRR 0.948 # 10
Hits@10 0.954 # 14
Hits@3 0.950 # 8
Hits@1 0.944 # 7
MR 412 # 14
training time (s) 1042 # 4
Node Classification YouTube LINE runtime (s) 70.09 # 1
Micro-F1@2% 40.61 # 1
Macro-F1@2% 33.69 # 1

Methods used in the Paper


METHOD TYPE
🤖 No Methods Found Help the community by adding them if they're not listed; e.g. Deep Residual Learning for Image Recognition uses ResNet