Adversarial Stein Training for Graph Energy Models
arXiv:2108.12982
Abstract
Learning distributions over graph-structured data is a challenging task with many applications in biology and chemistry. In this work we use an energy-based model (EBM) based on multi-channel graph neural networks (GNN) to learn permutation invariant unnormalized density functions on graphs. Unlike standard EBM training methods our approach is to learn the model via minimizing adversarial stein discrepancy. Samples from the model can be obtained via Langevin dynamics based MCMC. We find that this approach achieves competitive results on graph generation compared to benchmark models.
arXiv commentAppeared at Machine Learning for Molecules Workshop at NeurIPS 2020.https://ml4molecules.github.io
References in corpus (16)
- Statistical mechanics of complex networks
- Junction Tree Variational Autoencoder for Molecular Graph Generation
- Adversarially Learned Inference
- An O(m) Algorithm for Cores Decomposition of Networks
- Learning Deep Generative Models of Graphs
- LanczosNet: Multi-Scale Deep Graph Convolutional Networks
- Measuring Sample Quality with Stein's Method
- Knowledge Transfer for Out-of-Knowledge-Base Entities: A Graph Neural Network Approach
- DP-GAN: Diversity-Promoting Generative Adversarial Network for Generating Informative and Diversified Text
- Operator Variational Inference
- Latent Normalizing Flows for Discrete Sequences
- Permutation Invariant Graph Generation via Score-Based Generative Modeling
- Stein Neural Sampler
- Energy-based models for atomic-resolution protein conformations
- Kernel Stein Tests for Multiple Model Comparison
- Bosonic Random Walk Networks for Graph Learning