AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Exploring Forensic Dental Identification with Deep Learning

Yuan Liang, Weikun Han, Liang Qiu, Chen Wu, Yiting Shao, Kun Wang, Lei He

NeurIPS 2021 · Architectures and efficiency › Attention and transformers · approach: Token merging and masked pretraining · 6 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 transformer self-attention with average pooling (ablation)FoID ResNet-34 · Dental radiograph retrievalmAP: 0.3707 -> 0.5465Improved
Replace transformer self-attention with max pooling (ablation)FoID ResNet-34 · Dental radiograph retrievalmAP: 0.2519 -> 0.5465Improved
Replace transformer self-attention with gated attention (ablation)FoID Inception-ResNet · Dental radiograph retrievalmAP: 0.3741 -> 0.5962Improved
Replace transformer self-attention with FC layers (ablation)FoID Inception-ResNet · Dental radiograph retrievalmAP: 0.3007 -> 0.5962Improved
Use hard dental attention (FoID) instead of soft attention PGFAFoID ResNet-34 · Dental radiograph retrievalmAP: 0.4982 -> 0.5465Improved
Use domain-specific augmentation instead of basic augmentation (ablation)FoID and CNN baselines · Dental radiograph retrievalmAP: 3.28% to 15.32%Improved
Combine contrastive and id classification losses versus either alone (ablation)FoID ResNet-34 · Dental radiograph retrievalmAP: 0.4760/0.3196 -> 0.5394Improved
Combine contrastive and id classification losses versus either alone (ablation)FoID Inception-ResNet · Dental radiograph retrievalmAP: 0.4438/0.2808 -> 0.5394Improved
Add deep supervision loss to the two losses (ablation)FoID ResNet-34 · Dental radiograph retrievalmAP: 0.5394 -> 0.5465Improved
Add deep supervision loss to the two losses (ablation)FoID Inception-ResNet · Dental radiograph retrievalmAP: 0.5394 -> 0.5962Improved
Compare age group over 30 with age 30 or underFoID · Dental radiograph retrievalmAP: 0.6322 -> 0.5161No clear effect
Compare male with female patientsFoID · Dental radiograph retrievalmAP: 0.6353 -> 0.5296No clear effect

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