Step 1= Start with 1000 artificial neurons with 400 connections. use a pattern recognition paradigm. measure how many iterations to teach say 5 patterns and how many to converge on the pattern when presenting degraded patterns.
Step 2-n= at each step 10 connections are deleted and compensation (more on this later is applied). A new pattern is taught. And the number of iteration to converge on the old patterns as well as the new pattern. Untill 0 connections.
Repeat this for 2000,4000, 8000 neurons.
1 freelancer está ofertando em média $250 para este trabalho
Hi man. I could work for you as I have worked a lot on Hopfield networks category. Just tell me do you wanna use Matlab toolbox or manual codes? What ever I could do it.