On convex decision regions in deep network representations

Lenka Tětková (), Thea Brüsch, Teresa Dorszewski, Fabian Martin Mager, Rasmus Ørtoft Aagaard, Jonathan Foldager, Tommy Sonne Alstrøm and Lars Kai Hansen
Additional contact information
Lenka Tětková: Technical University of Denmark
Thea Brüsch: Technical University of Denmark
Teresa Dorszewski: Technical University of Denmark
Fabian Martin Mager: Technical University of Denmark
Rasmus Ørtoft Aagaard: Technical University of Denmark
Jonathan Foldager: Technical University of Denmark
Tommy Sonne Alstrøm: Technical University of Denmark
Lars Kai Hansen: Technical University of Denmark

Nature Communications, 2025, vol. 16, issue 1, 1-12

Abstract: Abstract Current work on human-machine alignment aims at understanding machine-learned latent spaces and their relations to human representations. We study the convexity of concept regions in machine-learned latent spaces, inspired by Gärdenfors’ conceptual spaces. In cognitive science, convexity is found to support generalization, few-shot learning, and interpersonal alignment. We develop tools to measure convexity in sampled data and evaluate it across layers of state-of-the-art deep networks. We show that convexity is robust to relevant latent space transformations and, hence, meaningful as a quality of machine-learned latent spaces. We find pervasive approximate convexity across domains, including image, text, audio, human activity, and medical data. Fine-tuning generally increases convexity, and the level of convexity of class label regions in pretrained models predicts subsequent fine-tuning performance. Our framework allows investigation of layered latent representations and offers new insights into learning mechanisms, human-machine alignment, and potential improvements in model generalization.

Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
https://www.nature.com/articles/s41467-025-60809-y Abstract (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:nat:natcom:v:16:y:2025:i:1:d:10.1038_s41467-025-60809-y

Ordering information: This journal article can be ordered from
https://www.nature.com/ncomms/

DOI: 10.1038/s41467-025-60809-y

Access Statistics for this article

Nature Communications is currently edited by Nathalie Le Bot, Enda Bergin and Fiona Gillespie

More articles in Nature Communications from Nature
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().