Abstract
The homogeneous form of the alternating projection neural network (APNN) performs as a content-addressable memory. We analyze and illustrate the characteristics and performance of the homogeneous APNN. Convergence of the iterative reconstruction becomes faster when the percentage of the clamped neurons, corresponding to known states, increases and the number of stored library vectors decreases. For a bipolar (±1) library, we demonstrate one-step convergence when the number of output neurons is sufficiently small. A new per-step minimization method for relaxation is introduced and is favorably contrasted in performance to other relaxation methods. We also propose a modified training procedure that requires neither a global norm operation nor division. Lastly, the noise characteristics of the APNN are examined and illustrated.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
