Our framework enables the development of the first fully data-driven solutions to active touch on top of learned models for object understanding.

V is constrained to vary along a limited set of dimensions.

Recent workhasshown that neural networks can exhibit an inductive bias that is often introduced via designing specific structures[4].

However,in3Dshapereconstruction,the complementaryfusionof visual and haptic modalities remains largely unexplored.

An important result is that for linear functionsf it is possible to recovers from observationsx up to certain symmetries, i.e., the model isidentifiable [8].

Let t beanother F analytic in t suchthat 0 = f0 and ( t) =( ft) forallt. Intherightfigurea volumepreservingtransfor applied (see Appendix C). 3: Smoothinvariantdeformations.