changed log.reg. cost function to be more explicit in n00b doc

doc/n00b.txt

@@ -169,11 +169,11 @@ training by simple gradient descent.
 
         stepsize = T.scalar('stepsize')  # a stepsize for gradient descent
 
-        # DECLARE SOME VARIABLES
+        # REGRESSION MODEL AND COSTS TO MINIMIZE
 
         prediction = T.softmax(T.dot(x, w) + b)
-
-        cost = T.sum(T.kl_multinomial(targ=y, pred=prediction)) + l2_coef * T.sum(T.sum(w*w))
+        cross_entropy = T.sum(y * T.log(prediction) + (1-y) * T.log(1.0 - prediction), axis=1)
+        cost = T.sum(cross_entropy) + l2_coef * T.sum(T.sum(w*w))
 
         # GET THE GRADIENTS NECESSARY TO FIT OUR PARAMETERS