AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Modelling Long Range Dependencies in $N$D: From Task-Specific to a General Purpose CNN

David M Knigge, David W. Romero, Albert Gu, Efstratios Gavves, Erik J Bekkers, Jakub Mikolaj Tomczak, Mark Hoogendoorn, Jan-jakob Sonke

ICLR 2023 · Architectures and efficiency › CNN design · approach: Filter design and trained-filter analysis · 5 citations (OpenAlex)

Read the paper

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 CCNN₆,₃₈₀ instead of treating the image as a flattened sequenceCCNN on LRA Pathfinder · LRA Pathfinder (2DPathfinder vs Pathfinder)accuracy: 91.51 -> 96.00Improved
Use CCNN₆,₃₈₀ instead of treating the image as a flattened sequenceCCNN on LRA Image · LRA Image (2DImage vs Image)accuracy: 88.90 -> 91.12Improved
Replace the FlexCNNBlock with the proposed CCNNBlock (ablation)CCNN₄,₁₄₀ · SpeechCommands-MFCCaccuracy: 94.38 -> 95.01Improved
Replace the FlexCNNBlock with the proposed CCNNBlock (ablation)CCNN₄,₁₄₀ · sCIFAR10accuracy: 85.54 -> 90.30Improved
Replace the FlexCNNBlock with the proposed CCNNBlock (ablation)CCNN₄,₁₄₀ · CIFAR10accuracy: 86.34 -> 92.78Improved
Replace the FlexCNNBlock with the proposed CCNNBlock (ablation)CCNN₆,₃₈₀ · SpeechCommands-MFCCaccuracy: 94.92 -> 95.01Improved
Replace the FlexCNNBlock with the proposed CCNNBlock (ablation)CCNN₆,₃₈₀ · sCIFAR10accuracy: 84.68 -> 93.08Improved
Replace the FlexCNNBlock with the proposed CCNNBlock (ablation)CCNN₆,₃₈₀ · CIFAR10accuracy: 92.24 -> 95.20Improved
Use the S4Block instead of the proposed CCNNBlock (ablation)CCNN₄,₁₄₀ · SpeechCommands-MFCCaccuracy: 95.01 -> 73.60Hurt
Use the S4Block instead of the proposed CCNNBlock (ablation)CCNN₄,₁₄₀ · sCIFAR10accuracy: 90.30 -> 53.25Hurt
Use the S4Block instead of the proposed CCNNBlock (ablation)CCNN₄,₁₄₀ · CIFAR10accuracy: 92.78 -> 60.06Hurt
Train CCNN₄,₁₄₀ on voxelized ModelNet10, then evaluate on point cloudsCCNN₄,₁₄₀ · ModelNet10 test setaccuracy: 90.64 -> 90.99Improved

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