AI Paper Atlas

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

Local Concept Embeddings for Analysis of Concept Distributions in Vision DNN Feature Spaces

Georgii Mikriukov, Gesina Schwalbe, Korinna Bade

IJCV 2025 · Label-efficient and robust learning › Recognition and long-tail · approach: Self-explaining and visual explanation networks · 7 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 zeros initialization instead of ones for the concept vector (ablation)LoCE · MS COCO 950 samples, YOLOv5 layer l3IoU: 0.21 -> 0.55Improved
use zeros initialization instead of ones for the concept vector (ablation)LoCE · MS COCO 950 samples, YOLOv5 layer l3optimization failures %: 73.2 -> 22.2Improved
use zeros initialization instead of ones (ablation)LoCE · MS COCO 950 samples, DETR layer l6optimization failures %: 96.6 -> 1.2Improved
use zeros instead of ones initialization (ablation)LoCE · MS COCO 950 samples, MobileNetV3 layer l1optimization failures %: 3.8 -> 0.8Improved
increase loss resolution from 50x50 to 200x200 (ablation)LoCE · MS COCO 950 samples, SWIN layer l3IoU: 0.07 -> 0.60Improved
increase loss resolution from 50x50 to 200x200 (ablation)LoCE · MS COCO 950 samples, SWIN layer l1optimization failures %: 99.4 -> 71.7Improved
increase loss resolution from 100x100 to 200x200 (ablation)LoCE · MS COCO 950 samples, EfficientNet layer l8IoU: 0.55 -> 0.52Hurt
increase loss resolution from 50x50 to 200x200 (ablation)LoCE · MS COCO 950 samples, MobileNetV3 layer l6IoU: 0.69 -> 0.67Hurt
use image-local LoCE instead of global Net2Vec vectorLoCE · MS COCO overall, DETR deep layerIoU: 0.56 -> 0.68Improved
use sub-global instead of global generalized vectorLoCE generalization · MS COCO overall, SWIN deep layerIoU: 0.14 -> 0.49Improved
use sparse 16-filter Net2Vec instead of dense Net2VecNet2Vec · MS COCO overall, DETR deep layerIoU: 0.56 -> 0.33Hurt
generalize LoCEs to a global vector instead of sub-global vectorsLoCE · MS COCO overall, DETR deep layerIoU: 0.61 -> 0.45Hurt

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