Abstract

As applied to the description of a Fourier-holography layout with phase conjugation in the correlation plane as the implementation of a two-layer neural network, this paper discusses two models of the advancement of hypotheses: linear regression of the conditions of a problem from knowledge, and inductive inference. The factors that influence the adequacy of the hypotheses generated for the conditions of a problem are determined and numerically investigated. It is shown that the adequacy of the hypotheses increases as the number of spatial degrees of freedom of the patterns that represent the conditions of the problem (the generalized frequency) increases; moreover, because of internal correlation (as an attribute of the information), increasing the size of the pattern influences the adequacy more effectively than does high-frequency filtering.

© 2013 Optical Society of America

PDF Article

References

You do not have subscription access to this journal. Citation lists with outbound citation links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access OSA Member Subscription

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access OSA Member Subscription