AI Paper Atlas

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

Rethinking Robust Representation Learning under Fine-Grained Noisy Faces

Bingqi Ma, Guanglu Song, Boxiao Liu, Yu Liu

ECCV 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 1 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
add class-specific sub-centers to ArcFace (ablation)ArcFace · IJB-C, Mixture of NoisesTAR@FAR=1e-4: 90.03 -> 93.76Improved
add sub-center merging (ablation)ESL (sub-center) · IJB-C, Mixture of NoisesTAR@FAR=1e-4: 93.76 -> 94.51Improved
add sub-center producing (ablation)ESL (sub-center + merging) · IJB-C, Mixture of NoisesTAR@FAR=1e-4: 94.51 -> 95.54Improved
add sub-center dropping (ablation)ESL (without dropping) · IJB-C, Mixture of NoisesTAR@FAR=1e-4: 95.54 -> 96.22Improved
use lambda4 = 3 instead of 1 for merging threshold (ablation)ESL · IJB-C, hyperparameter sweepTAR@FAR=1e-4: 95.07 -> 96.22Improved
use lambda1 = 2 instead of 1 (ablation)ESL · IJB-C, hyperparameter sweepTAR@FAR=1e-4: 95.73 -> 96.22Improved
use initial sub-centers M=3 instead of 1 (ablation)ESL · IJB-C, hyperparameter sweepTAR@FAR=1e-4: 96.03 -> 96.22Improved
increase initial sub-centers from 3 to 4 or 5 (ablation)ESL · IJB-C, hyperparameter sweepTAR@FAR=1e-4: 96.22 -> 96.20No clear effect
replace Sub-center ArcFace with ESLSub-center ArcFace M=3 · IJB-C, MS1MV0TAR@FAR=1e-4: 93.72 -> 96.23Improved
replace SKH with ESLSKH + ArcFace M=4 · IJB-C, Mixture of NoisesTAR@FAR=1e-4: 95.16 -> 96.22Improved
use Sub-center ArcFace on mixed noise instead of plain ArcFaceArcFace · IJB-C, Mixture of NoisesTAR@FAR=1e-4: 90.03 -> 89.32Hurt
train ESL online instead of posterior cleaning with retrainingSKH + ArcFace M=3↓1 · IJB-C, MS1MV0GPU hours: 128 -> 80Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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