AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Cross-Domain Facial Expression Recognition: A Unified Evaluation Benchmark and Adversarial Graph Learning

Tianshui Chen, Tao Pu, Hefeng Wu, Yuan Xie, Lingbo Liu, Liang Lin

TPAMI 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Geometric alignment and synthetic-to-real adaptation · 157 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 holistic-local feature co-adaptation module (vs holistic features only) (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 66.13 -> 59.68 (decrease of 6.45%)Improved
add two stacked GCNs (vs concatenating holistic-local features) (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 59.99 -> 66.13Improved
use separate intra- and inter-domain GCNs (vs one single GCN) (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 61.49 -> 66.13Improved
add inter-domain GCN to intra-domain GCN only (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 63.94 -> 66.13Improved
add intra-domain GCN to inter-domain GCN only (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 61.34 -> 66.13Improved
use per-class statistical distributions (vs dataset-level mean) (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 66.13 -> 63.03Improved
both per-iteration update and ten-epoch reclustering (vs update every iteration only) (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 63.63 -> 66.13Improved
both per-iteration update and ten-epoch reclustering (vs reclustering every ten epochs only) (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 62.65 -> 66.13Improved
manually defined adjacency initialization (vs random matrices) (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 59.48 -> 66.13Improved
manually defined adjacency initialization (vs all-one matrices) (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 62.81 -> 66.13Improved
fine-tune adjacency matrices jointly (vs fixed) (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 56.01 -> 66.13Improved
GCN fusion (vs transformer-based fusion) (ablation)AGRA · CD-FER, source RAF-DB, mean of 5 targets (ResNet-50)accuracy (%): 61.10 -> 66.13Improved

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