Variational Neural Belief
Cross-source consensus on Variational Neural Belief from 1 sources and 5 claims.
1 sources · 5 claims
How it works
Risks & contraindications
Highlighted claims
- VNB represents uncertainty over object pose and contact parameters with a Gaussian mixture model. — Variational Neural Belief Parameterizations for Robust Dexterous Grasping under Multimodal Uncertainty
- The belief parameterization uses mixture logits, component means, and log standard deviations. — Variational Neural Belief Parameterizations for Robust Dexterous Grasping under Multimodal Uncertainty
- VNB uses Gumbel-Softmax and location-scale reparameterization to make samples differentiable. — Variational Neural Belief Parameterizations for Robust Dexterous Grasping under Multimodal Uncertainty
- The article argues that differentiable Gaussian-mixture samples allow CVaR gradients to flow through belief parameters into action optimization. — Variational Neural Belief Parameterizations for Robust Dexterous Grasping under Multimodal Uncertainty
- VNB may need many mixture components for highly multimodal or complex contact posteriors. — Variational Neural Belief Parameterizations for Robust Dexterous Grasping under Multimodal Uncertainty