D-HYPR: Harnessing Neighborhood Modeling and Asymmetry Preservation for Digraph Representation Learning (2021-12-22T00:00:00.000000Z)

TL;DR

The resulting formalism, Digraph Hyperbolic Networks (D-HYPR) -- albeit conceptually simple -- generalizes to digraphs where cycles and non-transitive relations are common, and is applicable to multiple downstream tasks including node classification, link presence prediction, and link property prediction.

Abstract

Digraph Representation Learning (DRL) aims to learn representations for directed homogeneous graphs (digraphs). Prior work in DRL is largely constrained (e.g., limited to directed acyclic graphs), or has poor generalizability across tasks (e.g., evaluated solely on one task). Most Graph Neural Networks (GNNs) exhibit poor performance on digraphs due to the neglect of modeling neighborhoods and preserving asymmetry. In this paper, we address these notable challenges by leveraging hyperbolic collaborative learning from multi-ordered and partitioned neighborhoods, and regularizers inspired by socio-psychological factors. Our resulting formalism, Digraph Hyperbolic Networks (D-HYPR) -- albeit conceptually simple -- generalizes to digraphs where cycles and non-transitive relations are common, and is applicable to multiple downstream tasks including node classification, link presence prediction, and link property prediction. In order to assess the effectiveness of D-HYPR, extensive evaluations were performed across 8 real-world digraph datasets involving 21 prior techniques. D-HYPR statistically significantly outperforms the current state of the art. We release our code at https://github.com/hongluzhou/dhypr

Authors

Gerard de Melo

7 papers

Samuel S. Sohn

2 papers

M. Kapadia

3 papers

TL;DR

Abstract

Authors

References68 items

Directed Graph Embeddings in Pseudo-Riemannian Manifolds

Do Transformers Really Perform Bad for Graph Representation?

Fully Hyperbolic Neural Networks

Lorentzian Graph Convolutional Networks

A Hyperbolic-to-Hyperbolic Graph Convolutional Network

Directed Acyclic Graph Neural Networks

Graph Geometry Interaction Learning

Directional Graph Networks

Ultrahyperbolic Representation Learning

Low-Dimensional Hyperbolic Knowledge Graph Embeddings

Directed Graph Convolutional Network

SIGN: Scalable Inception Graph Neural Networks

Directed Acyclic Graph

Hyperbolic Graph Attention Network

Constant Curvature Graph Convolutional Networks

Hyperbolic Graph Neural Networks

Hyperbolic Graph Convolutional Neural Networks

Spectral-based Graph Convolutional Network for Directed Graphs

Skeleton-Based Action Recognition With Directed Graph Neural Networks

Disentangled Graph Convolutional Networks

Gravity-Inspired Graph Autoencoders for Directed Link Prediction

A survey on graph kernels

Simplifying Graph Convolutional Networks

Hyperbolic Disk Embeddings for Directed Acyclic Graphs

Graph Neural Networks: A Review of Methods and Applications

ATP: Directed Graph Embedding with Asymmetric Transitivity Preservation

Node Representation Learning for Directed Graphs

Predict then Propagate: Graph Neural Networks meet Personalized PageRank

Deep Graph Infomax

Learning Mixed-Curvature Representations in Product Spaces

Exploiting Edge Features in Graph Neural Networks.

Learning Continuous Hierarchies in the Lorentz Model of Hyperbolic Geometry

Relational inductive biases, deep learning, and graph networks

Hyperbolic Neural Networks

Representation Tradeoffs for Hyperbolic Embeddings

SIDE: Representation Learning in Signed Directed Networks

Hyperbolic Entailment Cones for Learning Hierarchical Embeddings

Graph Attention Networks

A Comprehensive Survey of Graph Embedding: Problems, Techniques, and Applications

Fast Network Embedding Enhancement via High Order Proximity Approximation

Attention is All you Need

Inductive Representation Learning on Large Graphs

Neural Embeddings of Graphs in Hyperbolic Space

Poincaré Embeddings for Learning Hierarchical Representations

Scalable Graph Embedding for Asymmetric Proximity

Geometric Deep Learning: Going beyond Euclidean data

Variational Graph Auto-Encoders

Semi-Supervised Classification with Graph Convolutional Networks

Asymmetric Transitivity Preserving Graph Embedding

node2vec: Scalable Feature Learning for Networks

Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Dimensionality reduction on Anchorgraph with an efficient Locality Preserving Projection

Adam: A Method for Stochastic Optimization

Principal Component Analysis (PCA)

DeepWalk: online learning of social representations

Weisfeiler-Lehman Graph Kernels

Hyperbolic Geometry of Complex Networks

On curvature and temperature of complex networks

A Brief History of Generative Models for Power Law and Lognormal Distributions

Birds of a Feather: Homophily in Social Networks

MagNet: A Magnetic Neural Network for Directed Graphs

Directed Graph Contrastive Learning

Digraph Inception Convolutional Networks

A Comprehensive Survey on Graph Neural Networks

Kipf and MaxWelling

BIRDS OF A FEATHER : Homophily in Social Networks

Challenge 1: Neighborhood Modeling Challenge 2: Asymmetry Preservation

Learning CIKM ’22, October 17–21, 2022, Atlanta, GA, USA

Field of Study

Journal Information

Name

Venue Information

Name

Type

URL

Alternate Names