Bernoulli Embeddings for Graphs

TL;DR

This paper introduces Bernoulli embeddings for graph nodes, enabling efficient binary representations that outperform traditional spectral and real-valued embeddings in graph retrieval tasks.

Contribution

It proposes a novel Bernoulli embedding model for graphs, using continuous optimization of biased coin flips to produce effective binary node representations.

Findings

Bernoulli embeddings outperform spectral and real-valued embeddings in ranking tasks.

The model reduces retrieval latency with competitive accuracy.

Embeddings are learned via continuous optimization of biased coin flips.

Abstract

Just as semantic hashing can accelerate information retrieval, binary valued embeddings can significantly reduce latency in the retrieval of graphical data. We introduce a simple but effective model for learning such binary vectors for nodes in a graph. By imagining the embeddings as independent coin flips of varying bias, continuous optimization techniques can be applied to the approximate expected loss. Embeddings optimized in this fashion consistently outperform the quantization of both spectral graph embeddings and various learned real-valued embeddings, on both ranking and pre-ranking tasks for a variety of datasets.