ML applied to structures: social network graphs, collaboration graphs, user interaction graphs, and more!
Many prediction problems can be phrased as inferences over local neighborhoods of graphs. The graph represents the interaction between entities. To predict a property of an entity, it is useful to consider the features of the entity itself, the structure of its local neighborhood, and the features of the entities in its local neighborhood.
We present an approach for applying machine learning directly to such graph neighborhoods, allowing predictions for individual nodes to be learned from examples, bypassing the step of creating and tuning an inference model or summarizing the neighborhoods via a fixed set of hand-crafted features. Our approach is based on a multi-level architecture built from Long Short-Term Memory neural nets (LSTMs); the LSTMs learn how to summarize the neighborhood from data. We demonstrate the effectiveness of the proposed technique on a synthetic example and on real-world data related to crowdsourced grading, Bitcoin transactions, and Wikipedia edit reversions.
This code enables the application of machine learning to graph neighborhoods, allowing predictions for graph nodes to be learned from examples. The code uses LSTMs to process the features of all nodes in the local neighborhood of a given target node, yielding a prediction for the target node itself. The code includes implementations of multi-level sequence learners, LSTMs, and deep NNs.
We are also making available the (rather unpolished) code used to generate the results for the paper "Learning from Graph Neighborhoods Using LSTMs".