Radical Edward Posted January 13, 2004 Posted January 13, 2004 arr, he be diggin' up the dead mateys.
Neurocomp2003 Posted January 14, 2004 Posted January 14, 2004 Neural nets are quite easy to make....but it depends on three things 1)the extense of your knowledge in mathematics 2)the size of the neural network you want 3)your willingness to sit there and play with constant parameters as you and NN's differ frmo CAs in that NN's deal with real or boolean values CAS discrete values and boolean nets deal with {-1,0,1} NN's use more statistical equations while booleans don't necessarily and CAs are just NN's learn to change by equations while CAs just change(transition table). ANNs can be software only... that is they are algorithms but there are research groups taht are trying to make ANNs hardware... actually a group in colorado i believe has acheived a silicon based , but also you get quantum NNs, molecular NNs and of course the cell on chip idea. But alot of this are still in preliminaries.. as per language MATLAB is the best simulator interms of people without coding experience where as once you learn matlab and have been able to programme some large nets you'd want to move to C++ for performance reasons anyways if your trying to get studnets to volunteer it is best to post your research idea at a school like CMU, as a undergrad/grad thesis project. Either way CMU is the best school to look at because they have the majority of top notch NN researchers there. THough every school has at least on eor two good ones...my old school had SUe Becker. who worked under G.Hinton
NavajoEverclear Posted January 14, 2004 Posted January 14, 2004 Sayonara³ said in post #26 :Intelligence is no longer with us.
Recommended Posts