Skip to main content
Uber logo

Schedule rides in advance

Reserve a rideReserve a ride

Schedule rides in advance

Reserve a rideReserve a ride
AI

Meta-Graph: Few-Shot Link Prediction Using Meta-Learning

May 29, 2020 / Global
Featured image for Meta-Graph: Few-Shot Link Prediction Using Meta-Learning
Figure 1: Our Meta-Graph architecture incorporates a graph signature which renders input to be processed by two GCN encoders. From there, the training graphs are decoded into a final product that provides parameter optimizations for future model training.
Figure 2. The convergence AUC results for the different fractions of training edges across data sets. With only 10 percent of edges, Meta-Graph outperforms all the other methods across all training datasets. As the percentage of training edges increases, performance of Meta-Graph is better in all but one setting of 30 percent training edges in FirstMM DB.
Figure 3. The AUC results with only five gradient updates for different fractions of training edges across differing data sets. Meta-Graph outperforms all the other methods across all training datasets except in one setting of using 20 percent of training edges in Ego-Aminer.
Ankit Jain

Ankit Jain

Ankit Jain is a former research scientist of Uber AI.

Piero Molino

Piero Molino

Piero is a Staff Research Scientist in the Hazy research group at Stanford University. He is a former founding member of Uber AI where he created Ludwig, worked on applied projects (COTA, Graph Learning for Uber Eats, Uber’s Dialogue System) and published research on NLP, Dialogue, Visualization, Graph Learning, Reinforcement Learning and Computer Vision.

Joey Bose

Joey Bose

Joey Bose is a PhD candidate at McGill University and the Montreal Institute for Learning algorithms working on broad problems involving graph representation learning, generative modeling of structured data and adversarial learning. He was a former intern at Uber AI.

William Hamilton

William Hamilton

William L. Hamilton is an Assistant Professor of Computer Science at McGill University, a member of the Mila - Quebec Artificial Intelligence Institute, and a CIFAR Canada AI Chair. His research focuses on graph representation learning and natural language processing. Will’s research has received several awards—-including a best paper award from the National Academy of Sciences—and his work has been featured in outlets such as Wired, The New York Times, and the BBC.

Posted by Ankit Jain, Piero Molino, Joey Bose, William Hamilton

Category: