Scalable Deep Generative Modeling for Sparse Graphs
This work addresses scalability for researchers and practitioners in domains like chemistry and biology who need to generate large sparse graphs efficiently.
The authors tackled the scalability problem of deep graph generative models by developing BiGG, an autoregressive model that exploits graph sparsity to reduce generation time complexity from O(n^2) to O((n+m)log n) and enables parallelized training with O(log n) synchronization, achieving better quality and scaling to much larger graphs than previous methods.
Learning graph generative models is a challenging task for deep learning and has wide applicability to a range of domains like chemistry, biology and social science. However current deep neural methods suffer from limited scalability: for a graph with $n$ nodes and $m$ edges, existing deep neural methods require $Ω(n^2)$ complexity by building up the adjacency matrix. On the other hand, many real world graphs are actually sparse in the sense that $m\ll n^2$. Based on this, we develop a novel autoregressive model, named BiGG, that utilizes this sparsity to avoid generating the full adjacency matrix, and importantly reduces the graph generation time complexity to $O((n + m)\log n)$. Furthermore, during training this autoregressive model can be parallelized with $O(\log n)$ synchronization stages, which makes it much more efficient than other autoregressive models that require $Ω(n)$. Experiments on several benchmarks show that the proposed approach not only scales to orders of magnitude larger graphs than previously possible with deep autoregressive graph generative models, but also yields better graph generation quality.