AI Paper Atlas

Research map / Large language models / Agents, code and math

Abstracting Concept-Changing Rules for Solving Raven’s Progressive Matrix Problems

Fan Shi, Bin Li, Xiangyang Xue

TPAMI 2026 · Large language models › Agents, code and math · approach: Inductive and abstract reasoning evaluation · 6 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 CRAB concept-wise generation without auxiliary supervision instead of TransformerCRAB · RAVEN averageselection accuracy: 59.8 -> 94.7Improved
use CRAB instead of Transformer on grid configurationCRAB · RAVEN 3x3Gridselection accuracy: 34.2 -> 90.2Improved
use CRAB instead of Transformer on inside-outside configurationCRAB · RAVEN O-ICselection accuracy: 14.5 -> 94.6Improved
use CRAB without auxiliary supervision instead of PrAE with distractors and rulesCRAB · RAVEN averageselection accuracy: 83.3 -> 94.7Improved
remove rule annotations from PrAE training (ablation)PrAE · RAVEN averageselection accuracy: 83.3 -> 13.6Hurt
test on I-RAVEN instead of RAVEN, selective solver SCLSCL · I-RAVEN vs RAVEN averageselection accuracy: 95.1 -> 85.7Hurt
test on I-RAVEN instead of RAVEN, CoPINetCoPINet · I-RAVEN vs RAVEN averageselection accuracy: 95.6 -> 15.2Hurt
test on I-RAVEN instead of RAVEN, CRABCRAB · I-RAVEN vs RAVEN averageselection accuracy: 94.7 -> 95.9No clear effect
use CRAB instead of selective SCL on I-RAVEN gridCRAB · I-RAVEN 3x3Gridselection accuracy: 42.5 -> 91.1Improved
increase target images and candidate set size in arbitrary-position generationCRAB · RAVEN GSAglobal selection accuracyNo clear effect
increase target images from 1 to 2 or moreTransformer · RAVEN GSAglobal selection accuracyHurt

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