AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Composable Sparse Subnetworks via Maximum-Entropy Principle

Francesco Caso, Samuele Fonio, Simone Monaco, Nicola Saccomanno, Fabrizio Silvestri

ICLR 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Sparse training from scratch · 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 ME loss instead of Quasi-MaxEnt lossShallow MLP, logit merge · Yeast pairwise merge |R|=2rewarded accuracy: 0.474 -> 0.565Improved
use ME loss instead of cross-entropyCNN, logit merge · FMNIST pairwise merge |R|=1rewarded accuracy: 0.633 -> 0.983Improved
use ME loss instead of cross-entropyShallow MLP, weight merge · FMNIST pairwise merge |R|=5rewarded accuracy: 0.434 -> 0.827Improved
merge in logit space instead of weight spaceDeep MLP with ME · HAR pairwise merge |R|=2rewarded accuracy: 0.870 -> 0.942Improved
add IMP before weight-space merging (ablation)Deep MLP with ME · MNIST and FMNIST pairwise mergerewarded accuracyImproved
add IMP before weight-space merging (ablation)Shallow MLP and CNN with ME · MNIST and FMNIST pairwise mergerewarded accuracyImproved
add IMP to ME training of single-class modules (ablation)Shallow MLP · MNIST single-class submodulerewarded accuracy: 0.998 -> 0.999No clear effect
add IMP (ablation)ResNet18 with ME · FMNIST and MNIST pairwise mergerewarded accuracyNo clear effect
re-estimate BatchNorm statistics on union of rewarded classes (ablation)VGG11 weight merge with IMP · FMNIST pairwise mergerewarded accuracyImproved
prune to 99% sparsity instead of 90% or 99.9% (ablation)Shallow MLP · MNIST complete mergerewarded accuracy and entropyImproved
use pruning ratio 0.6 instead of other ratios (ablation)CNN · MNIST and FMNIST pairwise mergerewarded accuracyImproved
use softmax temperature t=5 instead of 1, 3 or 7 (ablation)Shallow MLP with ME · MNIST complete mergeentropy and rewarded accuracyImproved

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