補足
結論から言って、単純なこのモデルで 98.25% が不可能では無いということは、非常に興味深い。
これは、単純なニューラルネットワークの限界に対しての誤認が大きい事を示唆していないだろうか？
認識率 100% は、相当の自信の表れ。 高価な GPU を使用していなくても十分テストは可能
本当のところ 100%は、printf の %.2f が原因、99.995以上は、100 として表示される。
浮動小数点を使用時に常に問題になる現象 -0の出現も含め。
使用するデータセットは、QMNIST の方がメンテナンスされていて良い可能性も追記。
https://github.com/facebookresearch/qmnist
使用した ニューラルネットワーク 構造
// ニューラルネットワーク構造
typedef struct {
float w1[IN_PUT_SIZE][HIDDEN_SIZE];
float b1[HIDDEN_SIZE];
float w2[HIDDEN_SIZE][OUTPUT_SIZE];
float b2[OUTPUT_SIZE];
} NeuralNetwork;
// 順伝播用バッファ
typedef struct {
float hidden[HIDDEN_SIZE];
float output[OUTPUT_SIZE];
} ForwardPass;
初期化時の初期値が非常に重要、この一択。
Loading training data ...
load_training_data_Extended = 60000
Epoch 1, Loss: 0.0091, Accuracy: 0.9969
Epoch 2, Loss: 0.0046, Accuracy: 0.9985
Epoch 3, Loss: 0.0031, Accuracy: 0.9993
Epoch 4, Loss: 0.0024, Accuracy: 0.9996
Epoch 5, Loss: 0.0021, Accuracy: 0.9997
Epoch 6, Loss: 0.0018, Accuracy: 0.9998
Epoch 7, Loss: 0.0016, Accuracy: 0.9998
Epoch 8, Loss: 0.0015, Accuracy: 0.9999
Epoch 9, Loss: 0.0014, Accuracy: 0.9999
Epoch 10, Loss: 0.0013, Accuracy: 0.9999
Epoch 11, Loss: 0.0012, Accuracy: 0.9999
Epoch 12, Loss: 0.0011, Accuracy: 0.9999
Epoch 13, Loss: 0.0010, Accuracy: 0.9999
Epoch 14, Loss: 0.0010, Accuracy: 0.9999
Epoch 15, Loss: 0.0009, Accuracy: 0.9999
Epoch 16, Loss: 0.0009, Accuracy: 0.9999
Epoch 17, Loss: 0.0009, Accuracy: 0.9999
Epoch 18, Loss: 0.0008, Accuracy: 1.0000
Epoch 19, Loss: 0.0008, Accuracy: 1.0000
Epoch 20, Loss: 0.0008, Accuracy: 1.0000
Epoch 21, Loss: 0.0007, Accuracy: 1.0000
Epoch 22, Loss: 0.0007, Accuracy: 1.0000
Epoch 23, Loss: 0.0007, Accuracy: 1.0000
Loading evaluation data ...
TEST_00000.7.png
7 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7:100.00% 8: 0.00%
9: 0.00%
TEST_00001.2.png 2 0: 0.00%
1: 0.00% 2:100.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_00002.1.png
1 0: 0.00% 1:100.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_00003.0.png
0 0:100.00% 1: 0.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_00004.4.png
4 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4:100.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_00005.1.png 1 0: 0.00%
1:100.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_00006.4.png
4 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4:100.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_00007.9.png 9 0: 0.00% 1:
0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9:100.00%
TEST_00008.5.png
5 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5: 99.97%
6: 0.03% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_00009.9.png 9 0: 0.00% 1:
0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9:100.00%
TEST_00010.0.png
0 0:100.00% 1: 0.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_00011.6.png
6 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6:100.00% 7: 0.00% 8: 0.00%
9: 0.00%
TEST_00012.9.png 9 0: 0.00%
1: 0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9:100.00%
TEST_00013.0.png
0 0:100.00% 1: 0.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_00014.1.png
1 0: 0.00% 1:100.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_00015.5.png
5 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.01% 4: 0.00% 5: 99.99%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_00016.9.png 9 0: 0.00% 1:
0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9:100.00%
TEST_00017.7.png
7 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7:100.00% 8: 0.00%
9: 0.00%
TEST_00018.3.png 3 0: 0.00%
1: 0.00% 2: 0.00% 3:100.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_00019.4.png
4 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4:100.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_00020.9.png 9 0: 0.00% 1:
0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9:100.00%
TEST_00021.6.png
6 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6:100.00% 7: 0.00% 8: 0.00%
9: 0.00%
TEST_00022.6.png 6 0: 0.00%
1: 0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6:100.00% 7: 0.00%
8: 0.00% 9: 0.00%
TEST_00023.5.png 5
0: 0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5:100.00% 6: 0.00%
7: 0.00% 8: 0.00% 9: 0.00%
TEST_00024.4.png
4 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 99.97% 5: 0.00%
6: 0.00% 7: 0.02% 8: 0.00% 9:
0.01%
TEST_00025.0.png 0 0:100.00% 1:
0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_00026.7.png
7 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7:100.00% 8: 0.00%
9: 0.00%
TEST_00027.4.png 4 0: 0.00%
1: 0.00% 2: 0.00% 3: 0.00%
4:100.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_00028.0.png
0 0:100.00% 1: 0.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_00029.1.png
1 0: 0.00% 1:100.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_00030.3.png
3 0: 0.00% 1: 0.00% 2:
0.00% 3:100.00% 4: 0.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_00031.1.png 1 0: 0.00% 1:
99.97% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.03% 8: 0.00% 9: 0.00%
TEST_00032.3.png
3 0: 0.00% 1: 0.00% 2:
0.00% 3:100.00% 4: 0.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_00033.4.png 4 0: 0.00% 1:
0.00% 2: 0.00% 3: 0.00% 4: 99.98%
5: 0.00% 6: 0.01% 7: 0.00% 8:
0.00% 9: 0.00%
TEST_00034.7.png 7 0:
0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7:100.00% 8: 0.00% 9: 0.00%
TEST_00035.2.png
2 0: 0.00% 1: 0.00% 2:100.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_00036.7.png
7 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7:100.00% 8: 0.00%
9: 0.00%
TEST_00037.1.png 1 0: 0.00%
1:100.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
:
TEST_09930.0.png
0 0:100.00% 1: 0.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09931.1.png
1 0: 0.00% 1:100.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09932.9.png
9 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7: 0.00% 8:
0.00% 9:100.00%
TEST_09933.2.png 2 0:
0.00% 1: 0.00% 2:100.00% 3: 0.00%
4: 0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_09934.8.png
8 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7: 0.00% 8:100.00%
9: 0.00%
TEST_09935.7.png 7 0: 0.00%
1: 0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:100.00%
8: 0.00% 9: 0.00%
TEST_09936.8.png 8
0: 0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8:100.00% 9: 0.00%
TEST_09937.2.png
2 0: 0.00% 1: 0.00% 2:100.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09938.6.png
6 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6:100.00% 7: 0.00% 8: 0.00%
9: 0.00%
TEST_09939.0.png 0 0:100.00%
1: 0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_09940.6.png
6 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6:100.00% 7: 0.00% 8: 0.00%
9: 0.00%
TEST_09941.5.png 5 0: 0.00%
1: 0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5:100.00% 6: 0.00% 7: 0.00%
8: 0.00% 9: 0.00%
TEST_09942.3.png 3
0: 0.00% 1: 0.00% 2: 0.00%
3:100.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09943.3.png
3 0: 0.00% 1: 0.00% 2:
0.00% 3:100.00% 4: 0.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_09944.3.png 3 0: 0.00% 1:
0.00% 2: 0.00% 3: 99.55% 4: 0.00%
5: 0.00% 6: 0.00% 7: 0.00% 8:
0.45% 9: 0.00%
TEST_09945.9.png 9 0:
0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9:100.00%
TEST_09946.1.png
1 0: 0.00% 1:100.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09947.4.png
4 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4:100.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_09948.0.png 0 0:100.00% 1:
0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_09949.6.png
6 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6:100.00% 7: 0.00% 8: 0.00%
9: 0.00%
TEST_09950.1.png 1 0: 0.00%
1:100.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_09951.0.png
0 0:100.00% 1: 0.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09952.0.png
0 0:100.00% 1: 0.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09953.6.png
6 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6:100.00% 7: 0.00% 8: 0.00%
9: 0.00%
TEST_09954.2.png 2 0: 0.00%
1: 0.00% 2:100.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_09955.1.png
1 0: 0.00% 1:100.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09956.1.png
1 0: 0.00% 1:100.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09957.7.png
7 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7:100.00% 8: 0.00%
9: 0.00%
TEST_09958.7.png 7 0: 0.00%
1: 0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:100.00%
8: 0.00% 9: 0.00%
TEST_09959.8.png 8
0: 0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8:100.00% 9: 0.00%
TEST_09960.4.png
4 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4:100.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_09961.6.png 6 0: 0.00% 1:
0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6:100.00% 7: 0.00%
8: 0.00% 9: 0.00%
TEST_09962.0.png 0
0:100.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09963.7.png
7 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7:100.00% 8: 0.00%
9: 0.00%
TEST_09964.0.png 0 0:100.00%
1: 0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_09965.3.png
3 0: 0.00% 1: 0.00% 2:
0.00% 3:100.00% 4: 0.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_09966.6.png 6 0: 0.00% 1:
0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6:100.00% 7: 0.00%
8: 0.00% 9: 0.00%
TEST_09967.8.png 8
0: 0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8:100.00% 9: 0.00%
TEST_09968.7.png
7 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7:100.00% 8: 0.00%
9: 0.00%
TEST_09969.1.png 1 0: 0.00%
1:100.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_09970.5.png
5 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:100.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_09971.2.png 2 0: 0.00% 1:
0.00% 2:100.00% 3: 0.00% 4: 0.00%
5: 0.00% 6: 0.00% 7: 0.00% 8:
0.00% 9: 0.00%
TEST_09972.4.png 4 0:
0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4:100.00% 5: 0.00% 6: 0.00%
7: 0.00% 8: 0.00% 9: 0.00%
TEST_09973.9.png
9 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7: 0.00% 8:
0.00% 9:100.00%
TEST_09974.4.png 4 0:
0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4:100.00% 5: 0.00% 6: 0.00%
7: 0.00% 8: 0.00% 9: 0.00%
TEST_09975.3.png
3 0: 0.00% 1: 0.00% 2:
0.00% 3: 56.04% 4: 0.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 43.95% 9: 0.00%
TEST_09976.6.png
6 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.01% 5:
0.00% 6: 99.99% 7: 0.00% 8: 0.00%
9: 0.00%
TEST_09977.4.png 4 0: 0.00%
1: 0.00% 2: 0.00% 3: 0.00%
4:100.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_09978.1.png
1 0: 0.00% 1:100.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09979.7.png
7 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7:100.00% 8: 0.00%
9: 0.00%
TEST_09980.2.png 2 0: 0.00%
1: 0.00% 2:100.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_09981.6.png
6 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6:100.00% 7: 0.00% 8: 0.00%
9: 0.00%
TEST_09982.5.png 5 0: 0.00%
1: 0.00% 2: 0.00% 3: 0.04% 4:
0.00% 5: 97.97% 6: 1.98% 7: 0.00%
8: 0.00% 9: 0.00%
TEST_09983.0.png 0
0:100.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09984.1.png
1 0: 0.00% 1:100.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09985.2.png
2 0: 0.00% 1: 0.00% 2:100.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09986.3.png
3 0: 0.00% 1: 0.00% 2:
0.00% 3:100.00% 4: 0.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_09987.4.png 4 0: 0.00% 1:
0.00% 2: 0.00% 3: 0.00% 4:100.00%
5: 0.00% 6: 0.00% 7: 0.00% 8:
0.00% 9: 0.00%
TEST_09988.5.png 5 0:
0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5:100.00% 6: 0.00%
7: 0.00% 8: 0.00% 9: 0.00%
TEST_09989.6.png
6 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6:100.00% 7: 0.00% 8: 0.00%
9: 0.00%
TEST_09990.7.png 7 0: 0.00%
1: 0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:100.00%
8: 0.00% 9: 0.00%
TEST_09991.8.png 8
0: 0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8:100.00% 9: 0.00%
TEST_09992.9.png
9 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6: 0.00% 7: 0.00% 8:
0.00% 9:100.00%
TEST_09993.0.png 0 0:100.00%
1: 0.00% 2: 0.00% 3: 0.00% 4:
0.00% 5: 0.00% 6: 0.00% 7:
0.00% 8: 0.00% 9: 0.00%
TEST_09994.1.png
1 0: 0.00% 1:100.00% 2: 0.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09995.2.png
2 0: 0.00% 1: 0.00% 2:100.00%
3: 0.00% 4: 0.00% 5: 0.00% 6:
0.00% 7: 0.00% 8: 0.00% 9: 0.00%
TEST_09996.3.png
3 0: 0.00% 1: 0.00% 2:
0.00% 3:100.00% 4: 0.00% 5: 0.00%
6: 0.00% 7: 0.00% 8: 0.00% 9:
0.00%
TEST_09997.4.png 4 0: 0.00% 1:
0.00% 2: 0.00% 3: 0.00% 4:100.00%
5: 0.00% 6: 0.00% 7: 0.00% 8:
0.00% 9: 0.00%
TEST_09998.5.png 5 0:
0.00% 1: 0.00% 2: 0.00% 3:
0.00% 4: 0.00% 5:100.00% 6: 0.00%
7: 0.00% 8: 0.00% 9: 0.00%
TEST_09999.6.png
6 0: 0.00% 1: 0.00% 2:
0.00% 3: 0.00% 4: 0.00% 5:
0.00% 6:100.00% 7: 0.00% 8: 0.00%
9: 0.00%
Evaluation error rate: 1.25% (125/10000) Number of training 60000
Total processing time: 229.83 seconds (Training: 224.54, Evaluation: 0.25)
real 3m49.868s
user 0m0.842s
sys 0m1.469s
$
Click here to download the ZIP file. NN_log.zip
誤認した、画像データの一覧。（明らかに、100％の認識が不可能であることの証明）
