AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

LaplacianFormer:Rethinking Linear Attention with Laplacian Kernel

Zhe Feng, Sen Lian, Changwei Wang, Muyang Zhang, Tianlong Tan, Rongtao Xu, Weiliang Meng, Xiaopeng Zhang

ICLR 2026 · Architectures and efficiency › Attention and transformers · approach: Linear attention approximations

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 Newton-Schulz iteration instead of conjugate gradient for kernel inverse (ablation)LaplacianFormer-Tiny · ImageNet-1Ktop-1 accuracy (%): 79.2 -> 81.4Improved
Use Newton-Schulz instead of CG (ablation)LaplacianFormer-Small · ImageNet-1Ktop-1 accuracy (%): 81.4 -> 83.8Improved
Set Laplacian kernel scale lambda to 4 instead of 0.5 (ablation)LaplacianFormer-Tiny · ImageNet-1Ktop-1 accuracy (%): 79.4 -> 81.4Improved
Increase kernel scale lambda from 4 to 8 (ablation)LaplacianFormer-Tiny · ImageNet-1Ktop-1 accuracy (%): 81.4 -> 78.5Hurt
Replace Gaussian kernel with Laplacian kernel in SOFT++ and Skyformer (ablation)SOFT++, Skyformer · ImageNet-1Kconvergence and accuracyImproved
Use depthwise-convolution landmark selection instead of average pooling (ablation)LaplacianFormer · ImageNet-1Ktop-1 accuracyNo clear effect
Use LaplacianFormer-Medium instead of SOFT++-MediumLaplacianFormer · ImageNet-1Ktop-1 accuracy (%): 83.7 -> 85.3Improved
Use LaplacianFormer-Small instead of SOFT++-Small as detector backboneLaplacianFormer · COCO Mask R-CNN 1xbox AP: 43.8 -> 45.8Improved
Use LaplacianFormer-Medium instead of SOFT++-Medium under RetinaNetLaplacianFormer · COCO RetinaNet 1xbox AP: 44.3 -> 47.2Improved
Use Laplacian kernel CUDA kernelsLaplacianFormer · Kernel and solver runtimeexecution timeImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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