An importance sampling approach to learning structural representations of shape

This paper addresses the problem of learning archetypal structural models from examples. This is done by providing a generative model for graphs where the distribution of observed nodes and edges is governed by a set of independent Bernoulli trials with parameters to be estimated, however, the correspondences between sample node and model nodes is not known and must be estimated from local structure. The parameters are estimated maximizing the likelihood of the observed graphs, marginalizing it over all possible node correspondences. This is done adopting an importance sampling approach to limit the exponential explosion of the set of correspondences. The approach is used to summarize the variation in two different structural abstraction of shape: Delaunay graph over a set of image features and shock graphs. The experiments show that the approach can be used to recognize structures belonging to a same class.

[1]  Narendra Ahuja,et al.  Extracting Subimages of an Unknown Category from a Set of Images , 2006, 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06).

[2]  Richard Sinkhorn.  A Relationship Between Arbitrary Positive Matrices and Doubly Stochastic Matrices , 1964 .

[3]  Alan L. Yuille,et al.  FORMS: A flexible object recognition and modelling system , 1996, International Journal of Computer Vision.

[4]  Nir Friedman,et al.  Being Bayesian About Network Structure. A Bayesian Approach to Structure Discovery in Bayesian Networks , 2004, Machine Learning.

[5]  Richard C. Wilson,et al.  Spectral Generative Models for Graphs , 2007, 14th International Conference on Image Analysis and Processing (ICIAP 2007).

[6]  Azriel Rosenfeld,et al.  3-D Shape Recovery Using Distributed Aspect Matching , 1992, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  Fritz Wysotzki,et al.  Central Clustering of Attributed Graphs , 2004, Machine Learning.

[8]  Harry G. Barrow,et al.  Subgraph Isomorphism, Matching Relational Structures and Maximal Cliques , 1976, Inf. Process. Lett..

[9]  Nir Friedman,et al.  Being Bayesian about Network Structure , 2000, UAI.

[10]  David A. Forsyth,et al.  Human Tracking with Mixtures of Trees , 2001, ICCV.

[11]  I. Beichl,et al.  Approximating the Permanent via Importance Sampling with Application to the Dimer Covering Problem , 1999 .

[12]  B KimiaBenjamin,et al.  Shapes, shocks, and deformations I , 1995 .

[13]  Ah Chung Tsoi,et al.  A self-organizing map for adaptive processing of structured data , 2003, IEEE Trans. Neural Networks.

[14]  J. Hammersley,et al.  Monte Carlo Methods , 1965 .

[15]  P. Erdös.  ASYMMETRIC GRAPHS , 2022 .

[16]  Ben Taskar,et al.  Learning Probabilistic Models of Relational Structure , 2001, ICML.

[17]  Benjamin B. Kimia,et al.  Shapes, shocks, and deformations I: The components of two-dimensional shape and the reaction-diffusion space , 1995, International Journal of Computer Vision.

[18]  David Maxwell Chickering,et al.  Learning Bayesian Networks: The Combination of Knowledge and Statistical Data , 1994, Machine Learning.

[19]  Horst Bunke,et al.  Graph Clustering Using the Weighted Minimum Common Supergraph , 2003, GbRPR.

[20]  Tyng-Luh Liu,et al.  Approximate tree matching and shape similarity , 1999, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[21]  Edwin R. Hancock,et al.  Learning shape-classes using a mixture of tree-unions , 2006, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[22]  Miguel Cazorla,et al.  Constellations and the Unsupervised Learning of Graphs , 2007, GbRPR.