AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Contrast with Reconstruct: Contrastive 3D Representation Learning Guided by Generative Pretraining

Zekun Qi, Runpei Dong, Guofan Fan, Zheng Ge, Xiangyu Zhang, Kaisheng Ma, Li Yi

ICML 2023 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Inductive bias and minimal representations · 31 citations (OpenAlex)

Read the paper

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
Guide cross-modal contrastive learning with masked reconstructionReCon · ScanObjectNN PB_T50_RSaccuracy: 90.63% vs. 82.48%Improved
Add reconstruction guidance to cross-modal contrastive pretraining (ablation)ReCon · ScanObjectNNaccuracy: 88.42 -> 90.18Improved
Add image contrastive supervision alongside reconstruction and text supervision (ablation)ReCon · ScanObjectNNaccuracy: 89.77 -> 90.63Improved
Add self-contrastive supervision to reconstruction, text, and image objectives (ablation)ReCon · ScanObjectNNaccuracy: 90.63 -> 89.98Hurt
Add stop-gradient to ReCon-block cross-attention (ablation)ReCon · ScanObjectNNaccuracy: 81.60 -> 90.63Improved
Add stop-gradient to ReCon-block cross-attention (ablation)ReCon · ModelNet40accuracy: 89.7 -> 94.1Improved
Replace Smooth ℓ1 with InfoNCE (ablation)ReCon contrastive objective · ScanObjectNNaccuracy: 90.63 -> 90.11Hurt
Replace Smooth ℓ1 with InfoNCE (ablation)ReCon contrastive objective · ModelNet40accuracy: 94.1 -> 93.8Hurt
Use paired rather than category-shuffled image data during pretraining (ablation)ReCon · ScanObjectNNaccuracy: 90.22 -> 90.63Improved
Use paired rather than category-shuffled image data during pretraining (ablation)ReCon · ModelNet40 zero-shotaccuracy: 60.4 -> 66.8Improved
Ensemble multiple text prompts (ablation)ReCon zero-shot classifier · ModelNet40top-1 accuracy: 60.6 -> 61.7Improved
Use Rotation rather than no augmentation (ablation)ReCon fine-tuning · ScanObjectNNaccuracy: 87.44 -> 90.63Improved

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