SnapE -- Training Snapshot Ensembles of Link Prediction Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shaban, Ali, Paulheim, Heiko
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913458776178688
author Shaban, Ali
Paulheim, Heiko
author_facet Shaban, Ali
Paulheim, Heiko
contents Snapshot ensembles have been widely used in various fields of prediction. They allow for training an ensemble of prediction models at the cost of training a single one. They are known to yield more robust predictions by creating a set of diverse base models. In this paper, we introduce an approach to transfer the idea of snapshot ensembles to link prediction models in knowledge graphs. Moreover, since link prediction in knowledge graphs is a setup without explicit negative examples, we propose a novel training loop that iteratively creates negative examples using previous snapshot models. An evaluation with four base models across four datasets shows that this approach constantly outperforms the single model approach, while keeping the training time constant.
format Preprint
id arxiv_https___arxiv_org_abs_2408_02707
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle SnapE -- Training Snapshot Ensembles of Link Prediction Models
Shaban, Ali
Paulheim, Heiko
Machine Learning
Artificial Intelligence
Snapshot ensembles have been widely used in various fields of prediction. They allow for training an ensemble of prediction models at the cost of training a single one. They are known to yield more robust predictions by creating a set of diverse base models. In this paper, we introduce an approach to transfer the idea of snapshot ensembles to link prediction models in knowledge graphs. Moreover, since link prediction in knowledge graphs is a setup without explicit negative examples, we propose a novel training loop that iteratively creates negative examples using previous snapshot models. An evaluation with four base models across four datasets shows that this approach constantly outperforms the single model approach, while keeping the training time constant.
title SnapE -- Training Snapshot Ensembles of Link Prediction Models
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2408.02707