Re: Bug in Neurolib 1.4.1

[Alex Lancaster]

>>>>> "KL" == Konrad Lang <address@hidden> writes:

KL> I just came over a part of Neurolib 1.4.1 that seems not to work
KL> as expected: in Hebb2Layer.m, method: - setFromString: (unsigned
KL> char *)_str The weights are initialized from the given string, but
KL> for the weights in the hidden layer the same offset in the string
KL> is taken ...

[...]

KL> the (->) should be replaced by: 

KL> foo += powers[j]*_str[(inpSize * hidSize)+i*BYTESWGT+j]; 

KL> to make sure that all the values are set from different locations
KL> out of the string ...

Thanks for the report and patch, Konrad!

Yes, you're right it appears that the weights between hidden and
output layers (Wgts2) are getting the same weights as those between
the input and hidden layers (Wgts1).  I'll investigate this and make
sure it's fixed in the next release.

 --- Alex

-- 
  Alex Lancaster           |  e-mail: address@hidden
  Swarm Program            |     web: http://www.santafe.edu/~alex
  Santa Fe Institute       |     tel: +1-(505) 984-8800 (ext 242)
-------------------------------------------------------------------

                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.