A generative adversarial strategy for modeling relation paths in knowledge base representation learning

Zia, Tehseen, Zahid, Usman and Windridge, David ORCID: https://orcid.org/0000-0001-5507-8516 (2019) A generative adversarial strategy for modeling relation paths in knowledge base representation learning. In: KR2ML - Knowledge Representation and Reasoning Meets Machine Learning Workshop, NeurIPS 2019, Thirty-third Conference on Neural Information Processing Systems, 09-14 Dec 2019, Vancouver, Canada. . [Conference or Workshop Item]

PDF - Final accepted version (with author's formatting)
Download (383kB) | Preview
[img] PDF - Draft pre-submission version (with author's formatting)
Restricted to Repository staff and depositor only

Download (414kB)


Enabling neural networks to perform multi-hop (mh) reasoning over knowledge bases (KBs) is vital for tasks such as question-answering and query expansion. Typically, recurrent neural networks (RNNs) trained with explicit objectives are used to model mh relation paths (mh-RPs). In this work, we hypothesize that explicit objectives are not the most effective strategy effective for learning mh-RNN reasoning models, proposing instead a generative adversarial network (GAN) based approach. The proposed model – mh Relation GAN (mh-RGAN) – consists of two networks; a generator $G$, and discriminator $D$. $G$ is tasked with composing a mh-RP and $D$ with discriminating between real and fake paths. During training, $G$ and $D$ contest each other adversarially as follows: $G$ attempts to fool $D$ by composing an indistinguishably invalid mh-RP given a head entity and a relation, while $D$ attempts to discriminate between valid and invalid reasoning chains until convergence. The resulting model is tested on benchmarks WordNet and FreeBase datasets and evaluated on the link prediction task using MRR and HIT@ 10, achieving best-in-class performance in all cases.

Item Type: Conference or Workshop Item (Poster)
Research Areas: A. > School of Science and Technology > Computer Science
Item ID: 28083
Notes on copyright: Rights remain with the authors.
Useful Links:
Depositing User: David Windridge
Date Deposited: 11 Nov 2019 09:52
Last Modified: 19 Jun 2021 03:45
URI: https://eprints.mdx.ac.uk/id/eprint/28083

Actions (login required)

View Item View Item


Activity Overview

Additional statistics are available via IRStats2.