Papers › Understanding Gaussian Attention Bias of Vision Transformers Using Effective Receptive Fields

Understanding Gaussian Attention Bias of Vision Transformers Using Effective Receptive Fields

8 May 2023arXiv:2305.04722archive 2025-07-28

Bum Jun Kim, Hyeyeon Choi, Hyeonah Jang, Sang Woo Kim

Vision transformers (ViTs) that model an image as a sequence of partitioned patches have shown notable performance in diverse vision tasks. Because partitioning patches eliminates the image structure, to reflect the order of patches, ViTs utilize an explicit component called positional embedding. However, we claim that the use of positional embedding does not simply guarantee the order-awareness of ViT. To support this claim, we analyze the actual behavior of ViTs using an effective receptive field. We demonstrate that during training, ViT acquires an understanding of patch order from the positional embedding that is trained to be a specific pattern. Based on this observation, we propose explicitly adding a Gaussian attention bias that guides the positional embedding to have the corresponding pattern from the beginning of training. We evaluated the influence of Gaussian attention bias on the performance of ViTs in several image classification, object detection, and semantic segmentation experiments. The results showed that proposed method not only facilitates ViTs to understand images but also boosts their performance on various datasets, including ImageNet, COCO 2017, and ADE20K.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Fine-Grained Image ClassificationImage ClassificationObject DetectionSemantic Segmentationimage-classificationobject-detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Fine-Grained Image Classification Caltech-101 ViT-S/16 (RPE w/ GAB) Top-1 Error Rate 9.798% #11 of 18 Archive leaderboard report
Fine-Grained Image Classification Stanford Dogs ViT-B/16 (RPE w/ GAB) Accuracy 90.185% #18 of 24 Archive leaderboard report
Image Classification ImageNet ViT-B/16 (RPE w/ GAB) Top 1 Accuracy 81.484% #639 of 1060 Archive leaderboard report
Image Classification Stanford Cars ViT-B/16 (RPE w/ GAB) Accuracy 93.743 #9 of 24 Archive leaderboard report
Image Classification Stanford Cars ViT-M/16 (RPE w/ GAB) Accuracy 83.89 #23 of 24 Archive leaderboard report
Object Detection COCO test-dev Swin-S (RPE w/ GAB) box mAP 48.23 #108 of 225 Archive leaderboard report
Semantic Segmentation ADE20K val Swin-S (RPE w/ GAB) mIoU 46.41 #70 of 95 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

AttentionLinear LayerMulti-Head AttentionSoftmaxVision Transformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections