AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

Bridging Functional and Representational Similarity via Usable Information

Antonio Almudévar, Alfonso Ortega

ICML 2026 · Probabilistic and scientific modeling › Molecules and proteins · approach: Probabilistic embeddings and relational inference · 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 bidirectional stitching instead of one stitcherstitching-based functional similarity · 58 vision models, 12,459 layer pairsaccuracy-ratio difference between directionsImproved
widen predictive family from orthogonal to orthogonal+scale to affine (ablation)representational similarity · 58 vision modelsrepresentational similarityImproved
compare CKA with usable-information similarityCKA · MLP layer pairsR2 of fit: R2=0.937Improved
compare RSA with usable-information similarityRSA · MLP layer pairsR2 of fit: R2=0.912Improved
compare SVCCA with usable-information similaritySVCCA · MLP layer pairsR2 of fit: R2=0.650Mixed
use coarse 20-superclass task instead of 100 fine classesfunctional similarity · CIFAR-100functional similarityImproved
high representational similarity as condition for functional similarityrepresentational similarity · 58 vision modelsP(accuracy ratio > 0.95)Improved
low representational similarity as condition for functional similarityrepresentational similarity · 58 vision modelsfunctional similarityNo 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