EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

On the Importance of Separation and Labelling on the Hypersphere

Martin Lindström iD, Ragnar Thobaben iD, Mikael Skoglund iD

Entropy · 01.08.2026 · 28 (8) · 870

DOI: 10.3390/e28080870 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Entropy, online erschienen am 01.08.2026. Verfasst von Martin Lindström und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Information Science and Engineering, KTH Royal Institute of Technology, 100 44 Stockholm, Sweden. Die Arbeit nennt 50 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Swedish Research Council.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Good representations strive for diversity among dissimilar features, often by mapping latent representations onto the hypersphere and enforcing separation. At the same time, the latent space structure should align with data semantics. While prior work propagates the need for separation, it is difficult to systematically analyse and determine the importance that feature cluster separation and semantic alignment have on downstream performance. In this paper, we address this gap in the understanding of hyperspherical clustering methods and provide new insights into which properties promote good performance. Firstly, we introduce a principled analysis framework which disentangles the importance of cluster separation and labelling on performance. Secondly, we give a full characterisation of the optimal cluster separation, both through theoretical analysis and practical schemes that achieve near-optimal separation. The results show that even though cluster separation contributes to performance, especially in low dimensions, performance gains obtained by matching the cluster labelling to the input data structure are more significant, and aligning cluster labelling with the underlying data structure can compensate for poor separation.

Schlagwörter

Stichwörter der Autor:innen

cluster labellingcluster separationhyperspherical clusteringprototypical learningrepresentation geometry

Autor:innen und Institutionen

  • Department of Information Science and Engineering, KTH Royal Institute of Technology, 100 44 Stockholm, Sweden.

ORCID-Profile: Martin Lindström, Ragnar Thobaben, Mikael Skoglund

Förderung

  • Swedish Research Council

Zitieren

APA

Lindström, M., Thobaben, R., & Skoglund, M. (2026). On the Importance of Separation and Labelling on the Hypersphere. Entropy, 28(8), 870. https://doi.org/10.3390/e28080870

Vancouver

Lindström M, Thobaben R, Skoglund M. On the Importance of Separation and Labelling on the Hypersphere. Entropy. 2026;28(8):870. doi:10.3390/e28080870

BibTeX
@article{Lindstrm2026,
  author = {Lindström, Martin and Thobaben, Ragnar and Skoglund, Mikael},
  title = {On the Importance of Separation and Labelling on the Hypersphere},
  journal = {Entropy},
  year = {2026},
  volume = {28},
  number = {8},
  pages = {870},
  doi = {10.3390/e28080870}
}
Kostenlos starten →