AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

von Mises-Fisher Loss: An Exploration of Embedding Geometries for Supervised Learning

Tyler R. Scott, Andrew C. Gallagher, Michael C. Mozer

ICCV 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Learning from unlabeled and positive-unlabeled data · 39 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use cosine distance at test time (ablation)Euclidean loss · Cars196 retrievalmAP@R: 21.3 -> 23.2Improved
use cosine distance at test time (ablation)Euclidean loss · SOP retrievalmAP@R: 39.7 -> 42.1Improved
use cosine distance at test time (ablation)Hyperbolic loss · SOP retrievalmAP@R: 41.0 -> 44.0Improved
use cosine distance at test time (ablation)Hyperbolic loss · Cars196 retrievalmAP@R: 22.9 -> 25.0Improved
use vMF loss instead of Spherical lossSpherical loss · CIFAR100 ECEECE (%): 21.8 -> 7.9Improved
use vMF loss instead of Euclidean lossEuclidean loss · CIFAR10 ECEECE (%): 8.8 -> 5.9Improved
apply temperature scaling post hoc (ablation)Euclidean loss · CIFAR100 ECEECE (%): 20.6 -> 2.2Improved
apply temperature scaling post hoc (ablation)Spherical loss · CIFAR100 ECEECE (%): 21.8 -> 10.8Improved
apply temperature scaling post hoc (ablation)vMF loss · CIFAR100 ECEECE (%): 7.9 -> 8.0No clear effect
use spherical geometry (Spherical) instead of EuclideanEuclidean loss · CIFAR100 accuracyaccuracy (%): 69.21 -> 70.57Improved
use spherical geometry (Spherical) instead of EuclideanEuclidean loss · CIFAR10 accuracyaccuracy (%): 94.13 -> 93.99No clear effect
use vMF loss instead of Spherical lossSpherical loss · SOP retrievalmAP@R: 44.3 -> 38.3Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy