AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Neuro-Fuzzy Concept Learning for Interpretable Large Multimodal Models

Ritik Mishra, Vanshika Gupta, M. Sajid, M. Tanveer

ICML 2026 · Vision-language and multimodal › Human and embodied modeling · approach: Concept bottleneck models

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 neuro-fuzzy concept learning instead of CoX-LMM Semi-NMFNeuro-FeX on LLaVA · COCO token BusCLIPScore: 0.621 -> 0.667Improved
use neuro-fuzzy concept learning instead of CoX-LMMNeuro-FeX-v0 on LLaVA · COCO token CatCLIPScore: 0.649 -> 0.593Hurt
use neuro-fuzzy concepts on other backbones with 30 diverse tokensNeuro-FeX-v1 on Qwen2.5-VL · 30 diverse tokensmean CLIPScore: 0.616 -> 0.643Improved
use K-means-initialized Gaussian membership (v1/v2) instead of SVD-initialized cosine membership (v0) (ablation)Neuro-FeX on LLaVA · COCO token Dog overlapoverlap score: 0.067 -> 0.430Hurt
use RBF initialization instead of random initialization (ablation)Neuro-FeX · reconstruction errorreconstruction error: 4.47 -> 0.43Improved
add fuzzy gating on top of RBF init (ablation)Neuro-FeX · reconstruction errorreconstruction error: 0.46 -> 0.43No clear effect
decrease Gaussian scaling parameter gamma (ablation)Neuro-FeX · Dog token, Qwen2 and LLaVACLIPScoreImproved
use Neuro-FeX instead of SAEs for monosemantic conceptsNeuro-FeX on SigLIP SoViT-400m · Monosemanticity scoremean MS: 0.610 -> 0.711Improved
use Neuro-FeX instead of Matryoshka SAE on Qwen3Neuro-FeX on Qwen3 · Monosemanticity scoremean MS: 0.549 -> 0.544No 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