Learning Universal Graph Neural Network Embeddings With Aid Of Transfer Learning

22 Sep 2019Saurabh VermaZhi-Li Zhang

Learning powerful data embeddings has become a center piece in machine learning, especially in natural language processing and computer vision domains. The crux of these embeddings is that they are pretrained on huge corpus of data in a unsupervised fashion, sometimes aided with transfer learning... (read more)

PDF Abstract

Evaluation Results from the Paper


 SOTA for Graph Classification on COLLAB (using extra training data)

     Get a GitHub badge
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK USES EXTRA
TRAINING DATA
COMPARE
Graph Classification COLLAB DUGNN Accuracy 84.20% # 1
Graph Classification D&D DUGNN Accuracy 82.40% # 2
Graph Classification ENZYMES DUGNN Accuracy 67.30% # 5
Graph Classification IMDb-B DUGNN Accuracy 78.70% # 2
Graph Classification IMDb-M DUGNN Accuracy 56.10% # 1
Graph Classification PROTEINS DUGNN Accuracy 81.70% # 1
Graph Classification PTC DUGNN Accuracy 74.7 # 3