Differences in the log likelihood of the classes are

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: sampled points increases. The complex model is less efficient than any other invariant model but its role is relevant within the classification procedure: thanks to its slow convergence, it is possible to state that any model having a log-likelihood worse than the complex model is not compatible with the sampled set. Moreover, when the number of sampled point tends to be infinite, the log-likelihood of the complex model approaches the value of the invariant class corresponding to the symmetries exhibited by the sampled surface. The reference parameter set Ω7 of the complex class is empty. C3 - Planar surface. The planar surface has been investigated in more detail to exploit the potentiality of the proposed approach. Table III reports the results from the ranking of the seven classes over the planar surface with different size of the sampled set. n=10 n=12 n=15 n=25 n=50 n=75 n=100 n=200 3 7 1 6 5 4 2 -5,26 -15,20 -15,43 -15,80 -16,84 -20,20 -26,97 3 6 7 1 5 4 2 -5,48 -5,49 -14,96 -15,27 -15,57 -16,05 -19,97 3 6 1 7 2 5 4...
View Full Document

Ask a homework question - tutors are online