AI Paper Atlas

Research map / Architectures and efficiency / CNN design

SONIC: Spectral Oriented Neural Invariant Convolutions

Gijs Joppe Moens, Regina Beets-Tan, Eduardo H P Pooch

ICLR 2026 · Architectures and efficiency › CNN design · approach: Steerable rotation-equivariant CNNs · 0 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 SonicNet instead of ConvNet under combined perturbationSynthShape · combined perturbation 20Dice: 0.62 -> 0.76Improved
Use SonicNet instead of ViT under 1.50 rescalingSynthShape · rescaling 1.50Dice: 0.68 -> 0.74Improved
Use SonicNet instead of NIFF under Gaussian noise 0.2SynthShape · Gaussian noise sigma 0.2Dice: 0.92 -> 0.79Hurt
Use SonicNet instead of ConvNet under 45 degree rotationSynthShape · rotation 45 degDice: 0.28 -> 0.24Hurt
Use SonicNet instead of ConvNet on HalliGalliHalliGalli · HalliGalliaccuracy: 0.42 -> 1.00Improved
Add inference noise sigma=0.1 to HalliGalli (SonicNet)SonicNet · HalliGalliaccuracy: 1.00 -> 0.86Hurt
Use SonicNet instead of NIFF under HalliGalli noiseHalliGalli · HalliGalli (noise sigma=0.1)accuracy: 0.56 -> 0.86Improved
Use SonicNet instead of nnU-Net ResEnc L on KiTS3D segmentation · KiTS DSCDSC: 88.98 -> 88.55Hurt
Use SonicNet instead of nnU-Net ResEnc L on ACDC3D segmentation · ACDC DSCDSC: 91.40 -> 92.02Improved
Use SonicNet instead of nnU-Net with fewer parameters3D segmentation · parametersparameters (M): 31.12 -> 2.59Improved
Use SonicNet instead of nnU-Net on PROMISexternal validation · PROMIS AUROCAUROC: 0.646 -> 0.687Improved
Use SonicNet instead of nnU-Net on Prostate158external validation · Prostate158 AUROCAUROC: 0.814 -> 0.841Improved

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