AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Sparling: End-to-End Spatial Concept Learning via Extremely Sparse Activations

Kavi Gupta, Osbert Bastani, Armando Solar-Lezama

ICLR 2026 · Architectures and efficiency › RNNs and normalization · approach: Differentiable optimization layers

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
lower target density delta (more extreme sparsity) during annealing (ablation)Sparling · DigitCircle, LaTeX-OCR, AudioMNISTSequencefalse positive error / false negative errorMixed
lower target density delta (more extreme sparsity) (ablation)Sparling · DigitCircle, LaTeX-OCR, AudioMNISTSequenceconfusion errorImproved
lower target density delta (more extreme sparsity) (ablation)Sparling · DigitCircle, LaTeX-OCR, AudioMNISTSequenceend-to-end errorHurt
add sparse bottleneck (Sparling vs Non-Sparse baseline)end-to-end model · DigitCircle, LaTeX-OCR, AudioMNISTSequenceend-to-end errorHurt
remove bottleneck, freeze motif model and retrain h-hat (Retrained) (ablation)Sparling · DigitCircle, LaTeX-OCR, AudioMNISTSequenceend-to-end errorImproved
increase density 2-3x above the extreme-sparsity level (ablation)Sparling · motif domainsconfusion errorHurt

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