AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

TT-Sparse: Learning Sparse Rule Models with Differentiable Truth Tables

Hans Farrell Soegeng, Sarthak Ketanbhai Modi, Thomas Peyrin

ICML 2026 · Learning theory and algorithms › Prediction and decision losses · approach: Active learning and subset selection · 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
learn fan-in with soft TopK mapping instead of FC + L0 gates (ablation)TT-Sparse · HeartAUC: .893 -> .931Improved
learn fan-in with soft TopK mapping instead of FC + L1 + prune (ablation)TT-Sparse · DiabetesAUC: .676 -> .829Improved
learn connectivity with W_map instead of magnitude TopK (ablation)TT-Sparse · DiabetesAUC: .716 -> .829Improved
use W_map instead of magnitude TopK (ablation)TT-Sparse · AbaloneR2: .152 -> .512Improved
use soft TopK mapping instead of FC + L0 gates (ablation)TT-Sparse · AbaloneR2: -.00 -> .512Improved
use soft TopK mapping instead of FC + L1 + prune (ablation)TT-Sparse · Yeastrule complexity: 96 -> 110Hurt
use closed-form sigmoid soft TopK instead of SinkhornTT-Sparse · Differentiable TopK benchmarkmemory (MB): 1{,}075.7 -> 3.4Improved
use closed-form soft TopK instead of perturbed maximizerTT-Sparse · Differentiable TopK benchmarkgradient variance: 200.3 / 22.1 -> 0.0Improved
add don't-care terms to QMC minimization (k=6) (ablation)TT-Sparse · per-node rule complexity, 28 datasetsliterals per node: 19.24 -> 7.13Improved
use soft TopK instead of slot-based softmax selection from DWN (ablation)TT-Sparse · 28 datasetsAUC/R2Improved
add skip connection from input features to classifier (ablation)TT-Sparse · 28 datasetsAUC/R2Improved
increase thermometer bits beyond 7 (ablation)TT-Sparse · 28 datasetsAUC/R2No clear effect

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