Timezone: »

 
Poster
Harmonic Decompositions of Convolutional Networks
Meyer Scetbon · Zaid Harchaoui

Tue Jul 14 10:00 AM -- 10:45 AM & Tue Jul 14 11:00 PM -- 11:45 PM (PDT) @
in Poster Session 4 »

We present a description of the function space and the smoothness class associated with a convolutional network using the machinery of reproducing kernel Hilbert spaces. We show that the mapping associated with a convolutional network expands into a sum involving elementary functions akin to spherical harmonics. The functional decomposition can be related to functional ANOVA decompositions in nonparametric statistics. Building off this functional characterization, we obtain statistical bounds which highlight an interesting trade-off between the approximation error and the estimation error.

Keywords: Deep Learning Theory · Kernel Methods · Non-parametric Methods · General Machine Learning Techniques

Author Information

Meyer Scetbon (CREST, ENSAE)
Zaid Harchaoui (University of Washington)

More from the Same Authors