AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

TokenMixup: Efficient Attention-guided Token-level Data Augmentation for Transformers

Hyeong Kyu Choi, Joonmyung Choi, Hyunwoo J. Kim

NeurIPS 2022 · Architectures and efficiency › Attention and transformers · approach: Transformer expressivity and learnability · 12 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 attention-based rather than gradient-based saliency detectionHTM on CCT · CIFAR-100accuracy: 83.28 -> 83.56Improved
Use attention-based rather than gradient-based saliency detectionHTM on CCT · CIFAR-100 saliency detectionAvg. Latency (ms): 236 -> 16Improved
Add sample difficulty assessment with ScoreNet (ablation)HTM on CCT · CIFAR-100accuracy: 83.56 -> 82.85Improved
Add Hungarian optimal assignment for mixup pairs (ablation)HTM on CCT · CIFAR-100accuracy: 83.56 -> 82.61Improved
Use saliency-guided token-level mixup instead of random token replacement (ablation)HTM on CCT · CIFAR-100accuracy: 83.56 -> 83.05Improved
Add saliency-based label reassignment (ablation)HTM on CCT · CIFAR-100accuracy: 83.56 -> 82.62Improved
Add Horizontal TokenMixupCCT-7/3x1 · CIFAR-100 with Input Mixup and Cutmixaccuracy: 82.87 -> 83.56Improved
Add Vertical TokenMixupCCT-7/3x1 · CIFAR-100 with Input Mixup and Cutmixaccuracy: 82.87 -> 83.54Improved
Add both Horizontal and Vertical TokenMixupCCT-7/3x1 · CIFAR-100 with Input Mixup and Cutmixaccuracy: 82.87 -> 83.57Improved
Use Manifold Mixup instead of token mixupCCT-7/3x1 · CIFAR-100 with Input Mixup and Cutmixaccuracy: 82.87 -> 80.08Hurt

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