실행 결과
Epoch 00001: val_loss improved from inf to 0.08178, saving model to ../data/model\MNIST_CNN.hdf5 ... (중략) ... Epoch 00008: val_loss improved from 0.04101 to 0.03858, saving model to ../data/model\MNIST_CNN.hdf5 ... (중략) ... Epoch 00023: val_loss did not improve from 0.03742 313/313 [==============================] - 2s 6ms/step - loss: 0.0341 - accuracy: 0.9917 Test Accuracy: 0.9917
그림 16-10 | 학습 진행에 따른 학습셋과 테스트셋의 오차 변화