AI Paper Atlas

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

Gaussian Affinity for Max-Margin Class Imbalanced Learning

Munawar Hayat, Salman Khan, Syed Waqas Zamir, Jianbing Shen, Ling Shao

ICCV 2019 · Label-efficient and robust learning › Recognition and long-tail · approach: Learning from unlabeled and positive-unlabeled data · 106 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
replace cost-sensitive CNN with affinity lossCoSen CNN · DIL skin lesion 5 classesaccuracy: 80.2 -> 91.1Improved
replace cost-sensitive CNN with affinity lossCoSen CNN · DIL skin lesion 10 classesaccuracy: 72.6 -> 80.3Improved
replace CoSen CNN with affinity lossCoSen CNN · Imbalanced MNIST (10% of even digits)accuracy: 98.4 -> 99.3Improved
replace CoSen CNN with affinity lossCoSen CNN · Imbalanced MNIST (25% of even digits)accuracy: 98.5 -> 99.5Improved
replace CoSen CNN with affinity lossCoSen CNN · Imbalanced MNIST (standard split)accuracy: 99.3 -> 99.6Improved
replace soft-max loss with affinity loss under label noisesoft-max · MNIST with 10-50% flipped labelserror rateImproved
replace soft-max with affinity loss across imbalance levelssoft-max · MNIST imbalance up to 1:0.025error rateImproved
use at least 5 prototypes per class instead of fewer (ablation)Affinity Loss · Imbalanced MNIST (10% of first five digits)error rate: best for m >= 5, m <= 5 relatively poorImproved
increase prototypes per class beyond 5 (ablation)Affinity Loss · Imbalanced MNIST (10% of first five digits)error rateNo clear effect
set sigma between 5 and 20 instead of very high values (ablation)Affinity Loss · Imbalanced MNIST (10% of first five digits)error rate: optimal sigma between 5 and 20Improved
use inverse distance similarity instead of Gaussian with squared L2 (ablation)Affinity Loss · Imbalanced MNIST (10% of first five digits)accuracy: 99.3 -> 99.1Hurt
use L1 distance with Gaussian similarity (ablation)Affinity Loss · Imbalanced MNIST (10% of first five digits)accuracy: fails to convergeHurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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