I have a mixed dataset with categorical features and text features, for categorical features I am using the approach as in image
After the stochastic Gradient Descent classifier I am not able to figure out what is actually being fed to Neural Network. Is it the probabilities or weights. The final outcome is the probability of class, classification problem with 100 classes.
The second question is either there are two inputs fed to NN or the concatenation of the inputs sent as 1 input to NN?