From: artificialneuralnetwork3000@gmail.com   
      
   Beginner, need help with Artificial Neural Network code, Thanks !   
   Options   
      
   Hello all !   
      
   I'm a real begginer in this field, and I need some help here, I read   
   a   
   lot about it on the net but I'm still a little confused.....   
      
      
   First of all I'll explain what I need to do - my input (that I'll   
   give   
   to the network) is a 30 frames sequence, which represent a short   
   movie. Each frame contains 100 x 100 pixels, and each pixel can be   
   black or white (0/1) , each movie ( = 30 frames) contains a black   
   background, and on it, 1-20 white pixels that are standing in place   
   or   
   moving around. I need to know if the input movie contains pixels that   
   are moving in a Round circle paths.   
      
      
   So, if all the white pixels in the input movie are standing in place,   
   or moving in straight lines, then the network should give 0 as   
   result,   
   But if even one white pixel is moving in a circles path, then the   
   network should give 1 on the output.   
      
      
   1. How should I do that? should I insert all the 30 frames to the   
   network at once, in parallel? (so my first layer should contain 100 x   
   100 x 30 pixels) or should I insert them in serial one by one, one   
   frame at a time?   
      
      
   2. I didn't understand what is the Bias that I should give to each   
   neuron unit, is it just a constant random number? what is it good   
   for?   
      
      
   3. Will it help if I'll connect the Output of some neuron units to   
   the   
   Input of the Previous layers neuron units?   
      
      
   4. How many layers should my network have?  3, 7, 20 ? more ?   
      
      
   5. I thought to do the learning process by using an evolutionary   
   method, that is, to create a population of let's say 500 networks,   
   and   
   let them evolve till I'll get the result that I want. Will it work or   
   is there a better way for doing that?   
      
      
   6. Should the formula of each neuron be a simple sum of all the   
   (Inputs * weights) + the Bias ?   
      
      
   Please give me any advice, Thanks !   
      
   [ comp.ai is moderated ... your article may take a while to appear. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)