AI Paper Atlas

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

A Non-Isotropic Probabilistic Take On Proxy-Based Deep Metric Learning

Michael Kirchhof, Karsten Roth, Zeynep Akata, Enkelejda Kasneci

ECCV 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 13 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
make ProxyNCA probabilistic (EL-nivMF)ProxyNCA · CUB200-2011R@1: 63.2 -> 64.8Improved
make ProxyNCA probabilistic (EL-nivMF)ProxyNCA · CARS196R@1: 78.8 -> 82.1Improved
make ProxyNCA probabilistic (EL-nivMF)ProxyNCA · SOPR@1: 76.2 -> 76.6Improved
add EL-nivMF as regularizer on ProxyAnchorProxyAnchor · CUB200-2011R@1: 64.4 -> 66.5Improved
add EL-nivMF as regularizer on ProxyAnchorProxyAnchor · CARS196R@1: 82.4 -> 83.6Improved
add EL-nivMF as regularizer on ProxyAnchorProxyAnchor · SOPR@1: 78.0 -> 78.2Improved
add non-isotropy to cosine distance (d_nivMF) (ablation)proxy DML with d_Cos · CUB200-2011R@1: +2.1ppImproved
add non-isotropy to cosine distance (d_nivMF) (ablation)proxy DML with d_Cos · CARS196R@1: +1.7ppImproved
additionally consider image norm (d_EL-nivMF over d_nivMF) (ablation)non-isotropic proxy DML · CUB200-2011R@1: +0.6ppImproved
additionally consider image norm (d_EL-nivMF over d_nivMF) (ablation)non-isotropic proxy DML · CARS196R@1: +0.3ppImproved
use L2 distance instead of cosine distance as retrieval metric (ablation)models trained with various distances · CUB200-2011 and CARS196R@1: -5.34ppHurt
drop the normalizer D(K) in the nivMF density (ablation)nivMF · empirical testsperformanceHurt

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