AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

Margin-Mix: Semi–Supervised Learning for Face Expression Recognition

Corneliu Florea, Mihai Badea, Laura Florea, Andrei Racoviteanu, Constantin Vertan

ECCV 2020 · Label-efficient and robust learning › Semi-supervised learning · approach: GAN-based semi-supervised learning · 28 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 sharpening temperature 0.25 instead of 0.5 (ablation)MarginMix · FER+ (2000 labels)accuracy (%): 55.35 -> 60.83Improved
add sharpening (T=0.25) versus none (ablation)MarginMix · FER+ (2000 labels)accuracy (%): 57.29 -> 60.83Improved
use soft self-labels instead of hard nearest-centroid labels (ablation)MarginMix · FER+ (2000 labels)accuracy (%): 51.87 -> 60.83Improved
use lambda_M=1 instead of 0.5 (ablation)MarginMix · FER+ (2000 labels)accuracy (%): 60.44 -> 60.83No clear effect
use N_aug 1024 instead of 512 (ablation)MarginMix · FER+ (2000 labels)accuracy (%): 60.68 -> 60.83No clear effect
add parameter EMA (ablation)MarginMix · FER+ (2000 labels)accuracy (%): 59.85 -> 60.83No clear effect
replace MixMatch with MarginMixMarginMix · FER+ (320 labels)accuracy (%): 45.60 -> 50.76Improved
replace MixMatch with MarginMixMarginMix · FER+ (4000 labels)accuracy (%): 70.91 -> 75.18Improved
replace MixMatch with MarginMixMarginMix · RAF-DB (4000 labels)accuracy (%): 65.24 -> 70.68Improved
replace MixMatch with MarginMixMarginMix · CIFAR-10 (250 labels)error (%): 11.80 -> 10.76Improved
replace MixMatch with MarginMixMarginMix · CIFAR-10 (1000 labels)error (%): 7.75 -> 8.33Hurt
replace MixMatch with MarginMixMarginMix · CIFAR-100 (10000 labels)error (%): 28.88 -> 29.12Hurt

Later papers that compared with this method

1 later accepted paper reports MarginMix as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

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