AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Reverse-Complement Equivariant Networks for DNA Sequences

Vincent Mallet, Jean-Philippe Vert

NeurIPS 2021 · Architectures and efficiency › CNN design · approach: Group symmetry equivariant networks · 4 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
set Irrep ratio a/(a+b) to 0.75 instead of other ratios (0 to 1) (ablation)Irrep · Binary TF tasks, validationAuROC: 0.75Improved
use k-mer length k=3 instead of k in {1,2,4} (ablation)Irrep and Regular · Binary TF tasks, validationAuROC: 3Improved
replace Standard with RCPS equivariant architectureCNN · Binary TF tasksAuROCImproved
replace RCPS with Best Equivariant (Irrep/Regular)RCPS · Binary TF tasks, full dataAuROCImproved
train on 1,000 sequences instead of full data; Best Equivariant vs Standard gapBest Equivariant · Binary TF tasksAuROC gap vs Standard: 1% vs 0.3%Improved
equivariant models vs Standard for convergence speedequivariant models · Binary TF tasksconvergence speedNo clear effect
ensemble of two Irrep models instead of single Best EquivariantBest Equivariant · Binary TF tasksAuROCImproved
post-hoc averaging or ensembling two Standard modelsStandard · Binary TF tasksAuROCImproved
post-hoc Standard vs Ensemble StandardStandard · Binary TF tasks CTCF, SPI1, MAXAuROCMixed
Ensemble Irrep vs post-hoc and Ensemble Standard at two predictionsEnsemble Irrep · Binary TF tasksAuROCImproved
Irrep (k=3, a/(a+b)=75%) vs RCPS on profile task, full dataIrrep · ChIP-nexus profiles (Oct4, Sox2, Nanog, Klf4), full dataSpearman correlationHurt
Irrep vs RCPS on profile task, 1,000 training sequencesIrrep · ChIP-nexus profiles, reducedSpearman correlationImproved

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