Optimal Transport for structured data with application on graphs

Abstract

This work considers the problem of computing distances between structured objects such as undirected graphs, seen as probability distributions in a specific metric space. We consider a new transportation distance (i.e. that minimizes a total cost of transporting probability masses) that unveils the geometric nature of the structured objects space. Unlike Wasserstein or Gromov-Wasserstein metrics that focus solely and respectively on features (by considering a metric in the feature space) or structure (by seeing structure as a metric space), our new distance exploits jointly both information, and is consequently called Fused Gromov-Wasserstein (FGW). After discussing its properties and computational aspects, we show results on a graph classification task, where our method outperforms both graph kernels and deep graph convolutional networks. Exploiting further on the metric properties of FGW, interesting geometric objects such as Fréchet means or barycenters of graphs are illustrated and discussed in a clustering context.

Cited by in corpus (32)

• Metrics for Graph Comparison: A Practitioner's Guide
• Domain Adaptation for Robust Workload Level Alignment Between Sessions and Subjects using fNIRS
• Sliced Gromov-Wasserstein
• Cross-domain Named Entity Recognition via Graph Matching
• Scalable Optimal Transport Methods in Machine Learning: A Contemporary Survey
• Generalized Spectral Clustering via Gromov-Wasserstein Learning
• Distance Metric Learning for Graph Structured Data
• Projection-Free Optimization on Uniformly Convex Sets
• On a linear fused Gromov-Wasserstein distance for graph structured data
• Integrating Dynamical Systems Modeling with Spatiotemporal scRNA-seq Data Analysis
• Gromov-Wasserstein Factorization Models for Graph Clustering
• A contribution to Optimal Transport on incomparable spaces
• LCS Graph Kernel Based on Wasserstein Distance in Longest Common Subsequence Metric Space
• Graph Matching Optimization Network for Point Cloud Registration
• Wasserstein Embedding for Graph Learning
• RMS-FlowNet++: Efficient and Robust Multi-Scale Scene Flow Estimation for Large-Scale Point Clouds
• On Transportation of Mini-batches: A Hierarchical Approach
• COPT: Coordinated Optimal Transport for Graph Sketching
• Unbalanced Optimal Transport through Non-negative Penalized Linear Regression
• Learning transport cost from subset correspondence
• CO-Optimal Transport
• Unsupervised Learning of Graph Hierarchical Abstractions with Differentiable Coarsening and Optimal Transport
• Learning Graphons via Structured Gromov-Wasserstein Barycenters
• Sketching Merge Trees for Scientific Data Visualization
• A Wasserstein Graph Distance Based on Distributions of Probabilistic Node Embeddings
• Stability of Entropic Wasserstein Barycenters and application to random geometric graphs
• Online Graph Dictionary Learning
• Using Optimal Transport as Alignment Objective for fine-tuning Multilingual Contextualized Embeddings
• Characterization of Gromov-type geodesics
• Aligning Time Series on Incomparable Spaces
• Learning Graphon Autoencoders for Generative Graph Modeling
• Graph embedding using multi-layer adjacent point merging model