Recursive interferometric representations

Classification requires building invariant representations relatively to groups of deformations that preserve signal classes. Recursive interferometry computes invariants with a cascade of complex wavelet transforms and modulus operators. The resulting representation is stable relatively to elastic deformations and provides invariant representations of stationary processes. It maps signals to a manifold which preserves signal discriminability.

[1]  U. Grenander,et al.  Structural Image Restoration through Deformable Templates , 1991 .

[2]  G LoweDavid,et al.  Distinctive Image Features from Scale-Invariant Keypoints , 2004 .

[3]  Andrea J. van Doorn,et al.  The Structure of Locally Orderless Images , 1999, International Journal of Computer Vision.

[4]  Yann LeCun,et al.  What is the best multi-stage architecture for object recognition? , 2009, 2009 IEEE 12th International Conference on Computer Vision.

[5]  Lorenzo Rosasco,et al.  On Invariance in Hierarchical Models , 2009, NIPS.