General physical scene understanding requires more than simply localizin...
Given a visual scene, humans have strong intuitions about how a scene ca...
Humans rely on touch and tactile sensing for a lot of dexterous manipula...
The world is filled with articulated objects that are difficult to deter...
We propose HyperDynamics, a dynamics meta-learning framework that condit...
We present neural architectures that disentangle RGB-D images into objec...
We propose a method to optimize the representation and distinguishabilit...
Tensor CANDECOMP/PARAFAC (CP) decomposition has wide applications in
sta...