Place: Large lecture room.
Affiliation: KIIS Research Center at Università Ca’ Foscari Venezia, Italy.
Graph-based representations have been used with considerable success in computer vision in the abstraction and recognition of object shape and scene structure due to its ability to capture the invariants of the recognition problem and thus offer a representation that is robust to nuisance transformations such as pose or articulation. Despite this, the methodology available for learning structural representations from sets of training examples is relatively limited.
In this talk I will present continuing work in Venice that aim at the definition of generative approaches for graph learning applied to the recognition of deformable shapes, where we assume an underlying manifold structure abstracted through the structural representation, and propose generative models of the variations of the intrinsic metric of the corresponding Laplace operator.