AI Paper Atlas

Research map / Label-efficient and robust learning / Noisy labels

Robot Learning in Homes: Improving Generalization and Reducing Dataset Bias

Abhinav Gupta, Adithyavairavan Murali, Dhiraj Prakashchand Gandhi, Lerrel Pinto

NeurIPS 2018 · Label-efficient and robust learning › Noisy labels · approach: Contrastive learning on noisy data · 60 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
Train on diverse home grasp data instead of Lab-Baxter dataRobust-Grasp · Real-LCA overall, unseen homesgrasp accuracy (%): 18.33 -> 62.08Improved
Train on diverse home grasp data instead of Lab-Baxter dataPatch-Grasp · Lab-Baxter held-outaccuracy (%): 76.9 -> 71.5No clear effect
Train on diverse home grasp data instead of Lab-LCA dataPatch-Grasp · Home-LCA held-outaccuracy (%): 56.5 -> 69.9Improved
Add latent noise modeling and marginalization (ablation)Patch-Grasp trained on Home-LCA · Real-Sawyergrasp accuracy (%): 56.25 -> 77.50Improved
Add latent noise modeling and marginalization (ablation)Patch-Grasp trained on Home-LCA · Lab-Baxter held-outaccuracy (%): 71.5 -> 75.2Improved
Add latent noise modeling and marginalization (ablation)Patch-Grasp trained on Home-LCA · Home-LCA held-outaccuracy (%): 69.9 -> 73.0Improved
Train with home data rather than lab dataRobust-Grasp · Lab-LCA held-outaccuracy (%): 71.2 -> 71.1No clear effect
Omit the initial GPN-only pretraining stage (ablation)Robust-Grasp · Home-LCA grasp traininggrasp prediction performanceHurt

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