Graph Neural Networks with Missing Node Features | Emanuele Rossi

Join the Learning on Graphs and Geometry Reading Group:
https://hannes-stark.com/logag-readin...

Paper “On the Unreasonable Effectiveness of Feature Propagation in Learning on Graphs with Missing Node Features”: https://arxiv.org/abs/2111.12128

Abstract: While Graph Neural Networks (GNNs) have recently become the de facto standard for modeling relational data, they impose a strong assumption on the availability of the node or edge features of the graph. In many real-world applications, however, features are only partially available; for example, in social networks, age and gender are available only for a small subset of users. We present a general approach for handling missing features in graph machine learning applications that is based on minimization of the Dirichlet energy and leads to a diffusion-type differential equation on the graph. The discretization of this equation produces a simple, fast and scalable algorithm which we call Feature Propagation. We experimentally show that the proposed approach outperforms previous methods on seven common node-classification benchmarks and can withstand surprisingly high rates of missing features: on average we observe only around 4% relative accuracy drop when 99% of the features are missing. Moreover, it takes only 10 seconds to run on a graph with ∼2.5M nodes and ∼123M edges on a single GPU.

Authors:  Emanuele Rossi, Henry Kenlay, Maria I. Gorinova, Benjamin Paul Chamberlain, Xiaowen Dong, Michael Bronstein

Twitter Hannes:   / hannesstaerk  
Twitter Dominique:   / dom_beaini  
Twitter Valence Discovery:   / valence_ai  

Reading Group Slack: https://logag.slack.com/ssb/redirect#...

Chapters
~

00:00 Intro
00:16 Speaker Introduction
01:51 Why do we care graphs and missing node features?
04:23 Graph Neural Networks
11:08 Learning with missing node features
14:33 Reconstruction of graphs
33:12 Feature Propagation Algorithm
52:23 Differences with Label Propagation
58:22 Feature Propagation is Fast and Scalable
1:00:17 When does Feature Propagation work?
1:02:02 Future Directions & Conclusions
1:04:28 Q+A