AI Paper Atlas

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

Geoclidean: Few-Shot Generalization in Euclidean Geometry

Joy Hsu, Jiajun Wu, Noah Goodman

NeurIPS 2022 · Vision-language and multimodal › Human and embodied modeling · approach: Abstract visual reasoning benchmarks · 3 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
human vs pretrained vision featuresImageNet-pretrained VGG16/RN50/InV3/ViT · Geoclidean 37 conceptsaverage accuracy: 0.91 -> 0.70Hurt
high-level instead of low-level features (ablation)ViT · Geoclidean 37 conceptsaverage accuracy: 0.55 -> 0.70Improved
high-level instead of low-level features (ablation)VGG16 · Geoclidean 37 conceptsaverage accuracy: 0.55 -> 0.60Improved
high-level instead of low-level features (ablation)ResNet50 · Geoclidean 37 conceptsaverage accuracy: 0.55 -> 0.53Hurt
high-level instead of low-level features (ablation)InceptionV3 · Geoclidean 37 conceptsaverage accuracy: 0.54 -> 0.60Improved
ViT high-level vs convolutional high-level featuresViT · Geoclidean 37 conceptsaverage accuracy: 0.60 -> 0.70Improved
Far negatives instead of Close negativeshumans · Geoclidean concept llcaccuracy: 0.6767 -> 0.9233Improved
Far negatives instead of Close negativeshumans · Geoclidean concept cccaccuracy: 0.6667 -> 0.8767Improved
Far instead of Close negativeshumans · Geoclidean rhomboidaccuracy: 0.9700 -> 0.9300Hurt
ViT high-level vs humans on circle conceptsViT · Geoclidean concept cccaccuracy: 0.77 -> 0.90Improved
ViT high-level vs humans on ccccViT · Geoclidean concept ccccaccuracy: 0.85 -> 0.95Improved
InceptionV3 high-level alignment with humansInceptionV3 · Geoclidean 74 tasksPearson correlation: 0.2575Improved

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