AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Formal Concept Lattices are Good Semantic Scaffolds for Concept-Based Learning

Deepika Vemuri, Sayanta Adhikari, Ankit Saha, Krishn Vishwas Kher, Vineeth N. Balasubramanian

ICML 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Prototype and concept dictionary models · 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
add iterative class-group refinement (ablation)FoCA CBM · ImageNet100accuracy: 88.19 -> 91.92Improved
add iterative class-group refinement (ablation)FoCA CBM · CIFAR100accuracy: 77.19 -> 79.7Improved
add iterative class-group refinement (ablation)FoCA CBM · AwA2cluster impurity: 0.5231 -> 0.5288No clear effect
use joint instead of independent training (ablation)FoCA CBM · ImageNet100accuracy: 77.54 -> 91.88Improved
use joint instead of sequential training (ablation)FoCA CBM · ImageNet100accuracy: 86.69 -> 91.88Improved
supervise with random class groups instead of lattice groups (ablation)FoCA CBM · ImageNet100accuracy: 91.36 -> 28.96Hurt
replace FCA lattice with GPT4-built hierarchy (ablation)FoCA CBM · ImageNet100/AwA2/CIFAR100accuracy: 91.88 -> 91.47 (ImageNet100)No clear effect
use FCA hierarchy instead of GPT4 hierarchy under 1% attribute removal (ablation)FoCA CBM · ImageNet100accuracy drop (concept leakage): FCA 91.88 -> 60.57; GPT4 91.47 -> 89.20Improved
use hierarchical FoCA with lattice scaffolding on ViTFoCA ViT · CIFAR100accuracy: 84.49 -> 86.65Improved
increase attribute loss weight alpha from 0.01 to 1 (ablation)FoCA CBM · CIFAR100accuracy: 79.36 -> 73.50Hurt
increase group loss weight beta from 0.01 to 1 (ablation)FoCA CBM · CIFAR100accuracy: 79.36 -> 65.09Hurt
stack all lattice attribute sets after backbone (naive) instead of staged supervision (ablation)FoCA CBM-N · ImageNet100cluster impurity: 0.573 -> 0.665Hurt

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