03 Nov 2022

AKBC.png
Aaron_Sim_500x500.jpg
Aaron Sim
Lead Machine Learning Researcher

BenevolentAI is proud to sponsor the AKBC conference, which will be held in London 3-5 November at the Barbican Centre.

Catch the BenevolentAI team at AKBC, where they will present their latest paper on Pseudo-Riemannian Embedding Models for Multi-Relational Graph Representations.

‍Link to the conference

Abstract

Pseudo-Riemannian Embedding Models for Multi-Relational Graph Representations

Authors: Saee Gopal Paliwal, Angus Brayne, Benedek Fabian, Maciej Wiatrak, Aaron Sim.

In this paper we generalize single-relation pseudo-Riemannian graph embedding models to multi-relational networks, and show that the typical approach of encoding relations as manifold transformations translates from the Riemannian to the pseudo-Riemannian case. In addition we construct a view of relations as separate spacetime submanifolds of multi-time manifolds, and consider an interpolation between a pseudo-Riemannian embedding model and its Wick-rotated Riemannian counterpart. We validate these extensions in the task of link prediction, focusing on flat Lorentzian manifolds, and demonstrate their use in both knowledge graph completion and knowledge discovery in a biological domain.

Link to the paper


Back to events