Does Dlib's MLP implementation have any restriction?
问题 I am training a simple MLP (8 units in the input layer, 50 units in the hidden layer) using C++ DLib with the following 89 samples as training data: float template_user[89][8]={ {0.083651,0.281587,0.370476,0.704444,0.253865,0.056415,0.002344,0.465187}, {0.142540,0.272857,0.376032,0.740952,0.591501,0.227614,0.000000,0.832224}, {0.095625,0.258750,0.447500,0.779792,0.449932,0.000964,0.035104,0.606591}, {0.115208,0.181250,0.478750,0.797083,0.491855,0.015824,0.011652,0.649632}, {0.107436,0.166026