---
_id: '62007'
abstract:
- lang: eng
  text: "Ensemble methods are widely employed to improve generalization in machine
    learning. This has also prompted the adoption of ensemble learning for the knowledge
    graph embedding (KGE) models in performing link prediction. Typical approaches
    to this end train multiple models as part of the ensemble, and the diverse predictions
    are then averaged. However, this approach has some significant drawbacks. For
    instance, the computational overhead of training multiple models increases latency
    and memory overhead. In contrast, model merging approaches offer a promising alternative
    that does not require training multiple models. In this work, we introduce model
    merging, specifically weighted averaging, in\r\nKGE models. Herein, a running
    average of model parameters from a training epoch onward is maintained and used
    for predictions. To address this, we additionally propose an approach that selectively
    updates the running average of the ensemble model parameters only when the generalization
    performance improves on a validation dataset. We evaluate these two different
    weighted averaging approaches on link prediction tasks, comparing the state-of-the-art
    benchmark ensemble approach. Additionally, we evaluate the weighted averaging
    approach considering literal-augmented KGE models and multi-hop query answering
    tasks as well. The results demonstrate that the proposed weighted averaging approach
    consistently improves performance across diverse evaluation settings."
author:
- first_name: Rupesh
  full_name: Sapkota, Rupesh
  id: '89326'
  last_name: Sapkota
- first_name: Caglar
  full_name: Demir, Caglar
  last_name: Demir
- first_name: Arnab
  full_name: Sharma, Arnab
  last_name: Sharma
- first_name: Axel-Cyrille
  full_name: Ngonga Ngomo, Axel-Cyrille
  last_name: Ngonga Ngomo
citation:
  ama: 'Sapkota R, Demir C, Sharma A, Ngonga Ngomo A-C. Parameter Averaging in Link
    Prediction. In: <i>Proceedings of the Thirteenth International Conference on Knowledge
    Capture(K-CAP 2025)</i>. ACM; 2025. doi:<a href="https://doi.org/10.1145/3731443.3771365">https://doi.org/10.1145/3731443.3771365</a>'
  apa: Sapkota, R., Demir, C., Sharma, A., &#38; Ngonga Ngomo, A.-C. (2025). Parameter
    Averaging in Link Prediction. <i>Proceedings of the Thirteenth International Conference
    on Knowledge Capture(K-CAP 2025)</i>. Knowledge Capture Conference 2025, Dayton,
    OH, USA. <a href="https://doi.org/10.1145/3731443.3771365">https://doi.org/10.1145/3731443.3771365</a>
  bibtex: '@inproceedings{Sapkota_Demir_Sharma_Ngonga Ngomo_2025, place={Dayton, OH,
    USA}, title={Parameter Averaging in Link Prediction}, DOI={<a href="https://doi.org/10.1145/3731443.3771365">https://doi.org/10.1145/3731443.3771365</a>},
    booktitle={Proceedings of the Thirteenth International Conference on Knowledge
    Capture(K-CAP 2025)}, publisher={ACM}, author={Sapkota, Rupesh and Demir, Caglar
    and Sharma, Arnab and Ngonga Ngomo, Axel-Cyrille}, year={2025} }'
  chicago: 'Sapkota, Rupesh, Caglar Demir, Arnab Sharma, and Axel-Cyrille Ngonga Ngomo.
    “Parameter Averaging in Link Prediction.” In <i>Proceedings of the Thirteenth
    International Conference on Knowledge Capture(K-CAP 2025)</i>. Dayton, OH, USA:
    ACM, 2025. <a href="https://doi.org/10.1145/3731443.3771365">https://doi.org/10.1145/3731443.3771365</a>.'
  ieee: 'R. Sapkota, C. Demir, A. Sharma, and A.-C. Ngonga Ngomo, “Parameter Averaging
    in Link Prediction,” presented at the Knowledge Capture Conference 2025, Dayton,
    OH, USA, 2025, doi: <a href="https://doi.org/10.1145/3731443.3771365">https://doi.org/10.1145/3731443.3771365</a>.'
  mla: Sapkota, Rupesh, et al. “Parameter Averaging in Link Prediction.” <i>Proceedings
    of the Thirteenth International Conference on Knowledge Capture(K-CAP 2025)</i>,
    ACM, 2025, doi:<a href="https://doi.org/10.1145/3731443.3771365">https://doi.org/10.1145/3731443.3771365</a>.
  short: 'R. Sapkota, C. Demir, A. Sharma, A.-C. Ngonga Ngomo, in: Proceedings of
    the Thirteenth International Conference on Knowledge Capture(K-CAP 2025), ACM,
    Dayton, OH, USA, 2025.'
conference:
  end_date: 2025-12-10
  location: Dayton, OH, USA
  name: Knowledge Capture Conference 2025
  start_date: 2025-12-10
date_created: 2025-10-28T10:02:40Z
date_updated: 2025-12-04T09:15:07Z
ddc:
- '000'
department:
- _id: '574'
doi: https://doi.org/10.1145/3731443.3771365
file:
- access_level: open_access
  content_type: application/pdf
  creator: rupezzz
  date_created: 2025-10-28T10:02:13Z
  date_updated: 2025-10-28T10:02:13Z
  file_id: '62008'
  file_name: public.pdf
  file_size: 837462
  relation: main_file
file_date_updated: 2025-10-28T10:02:13Z
has_accepted_license: '1'
keyword:
- Knowledge Graphs
- Embeddings
- Ensemble Learning
language:
- iso: eng
main_file_link:
- url: https://papers.dice-research.org/2025/KCAP_ASWA/public.pdf
oa: '1'
place: Dayton, OH, USA
project:
- _id: '285'
  name: SAIL - Nachhaltiger Lebenszyklus von intelligenten soziotechnischen Systemen
publication: Proceedings of the Thirteenth International Conference on Knowledge Capture(K-CAP
  2025)
publisher: ACM
status: public
title: Parameter Averaging in Link Prediction
type: conference
user_id: '89326'
year: '2025'
...
