EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Enhancing deep hashing with graph filters and autoencoder-based embeddings

Sooin Kim, Kyungtae Kim, Seonghyun Min, Wongyum Kim, Doosung Hwang, Donghoon Kim

PeerJ Computer Science · 04.03.2026 · 12 · e3644

DOI: 10.7717/peerj-cs.3644 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PeerJ Computer Science, online erschienen am 04.03.2026. Verfasst von Sooin Kim und 5 weiteren Autor:innen. Die Arbeit nennt 50 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Korea Internet Security Agency (KISA) Grant funded by Korea Government, Research Development for Right to be forgotten in Video Contents provided by Limited Online Service Providers.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep hashing has emerged as an efficient and robust solution for image retrieval through representation learning. However, convolutional neural network (CNN)-based hashing methods are constrained by their reliance on grid structures, limiting their capacity to model complex or unstructured data relationships. This article proposes a novel deep hashing model, Hashing by Autoencoder with graph convolutional networks (HAGCN), that integrates transfer learning–based visual embeddings, obtained via an autoencoder, with graph convolutional networks (GCNs). The model dynamically constructs local subgraphs from the output of a transfer model, enabling the learning of both global and local structural relationships through the graph Laplacian. A GCN layer is employed to effectively capture local topologies in unstructured data, enhancing both representation quality and learning efficiency through parameter sharing and transfer learning. Experiments conducted on the evaluation datasets demonstrate that the proposed method outperforms existing CNN-based and GCN-based deep hashing approaches. Furthermore, the analysis of various GCN filters under the proposed framework offers valuable insights into filter selection for deep hashing. In terms of average performance improvement compared to the comparison models, HAGCN shows an increase of about 7.4% compared to the minimum performance and about 2.7% compared to the maximum performance. Ultimately, GCN filters contribute to structural preservation and improved expressiveness, while the combination of dynamic graph construction and transfer learning facilitates the generation of compact, robust hash codes from high-dimensional image data.

Förderung

  • Korea Internet Security Agency (KISA) Grant funded by Korea Government
  • Research Development for Right to be forgotten in Video Contents provided by Limited Online Service Providers

Zitieren

APA

Kim, S., Kim, K., Min, S., Kim, W., Hwang, D., & Kim, D. (2026). Enhancing deep hashing with graph filters and autoencoder-based embeddings. PeerJ Computer Science, 12, e3644. https://doi.org/10.7717/peerj-cs.3644

Vancouver

Kim S, Kim K, Min S, Kim W, Hwang D, Kim D. Enhancing deep hashing with graph filters and autoencoder-based embeddings. PeerJ Computer Science. 2026;12:e3644. doi:10.7717/peerj-cs.3644

BibTeX
@article{Kim2026,
  author = {Kim, Sooin and Kim, Kyungtae and Min, Seonghyun and Kim, Wongyum and Hwang, Doosung and Kim, Donghoon},
  title = {Enhancing deep hashing with graph filters and autoencoder-based embeddings},
  journal = {PeerJ Computer Science},
  year = {2026},
  volume = {12},
  pages = {e3644},
  doi = {10.7717/peerj-cs.3644}
}
Kostenlos starten →