Input layer 400, single hidden layer, hidden layer 100, the output layer is 15
Regular item 0.5
Iteration number 1000 (if reduce the number of iterations, also will make the two lower accuracy)
Fitting, the resulting neural network training set accuracy 100%, test set by 94% and increases the lambda results in the decrease of both
I hope both are about 98%, can you tell me how to adjust the cords?