Imports¶
In [54]:
%%capture
!pip install keras
!pip install tensorflow
In [55]:
import keras
import tensorflow
from keras import layers
from sklearn.datasets import load_breast_cancer
import pandas as pd
from sklearn.preprocessing import MinMaxScaler
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn.utils import shuffle
from sklearn.metrics import confusion_matrix
In [56]:
import warnings
warnings.filterwarnings("ignore")
Importing Wisconsin Breast Cancer Dataset from sklearn¶
In [57]:
data = load_breast_cancer()
print(data.keys())
dict_keys(['data', 'target', 'frame', 'target_names', 'DESCR', 'feature_names', 'filename', 'data_module'])
In [58]:
X = data.data
y = data.target
Creating two dataframes (features and respective targets) for EDA¶
In [59]:
features = pd.DataFrame(X, columns = data.feature_names)
targets = pd.DataFrame(y, columns = ['target'])
In [60]:
features.columns
Out[60]:
Index(['mean radius', 'mean texture', 'mean perimeter', 'mean area',
'mean smoothness', 'mean compactness', 'mean concavity',
'mean concave points', 'mean symmetry', 'mean fractal dimension',
'radius error', 'texture error', 'perimeter error', 'area error',
'smoothness error', 'compactness error', 'concavity error',
'concave points error', 'symmetry error', 'fractal dimension error',
'worst radius', 'worst texture', 'worst perimeter', 'worst area',
'worst smoothness', 'worst compactness', 'worst concavity',
'worst concave points', 'worst symmetry', 'worst fractal dimension'],
dtype='object')
In [61]:
#features.head().T
#targets.isnull().sum()
#targets.value_counts()
#features.info()
In [62]:
count_m = targets[targets['target'] == 1].shape[0]
count_b = targets[targets['target'] == 0].shape[0]
percentage_m = (count_m/targets.shape[0]) * 100
percentage_b = (count_b/targets.shape[0]) * 100
In [63]:
#for column in features.columns:
#plt.figure()
#sns.histplot(features[column], kde=True, color='skyblue')
#plt.title(f'Distribution of {column}', fontsize=16)
#plt.xlabel('Value', fontsize=14)
#plt.ylabel('Frequency', fontsize=14)
#plt.grid(True)
#plt.show()
EDA¶
In [64]:
features.shape
Out[64]:
(569, 30)
In [65]:
g = sns.FacetGrid(features.melt(), col='variable', col_wrap=4, sharex=False, sharey=False, height=3)
g.map(sns.kdeplot, 'value', fill=True)
g.set_titles('Density Plot: {col_name}')
plt.tight_layout()
plt.show()
In [66]:
correlation_matrix = features.corr()
sns.heatmap(correlation_matrix, annot=False, cmap='coolwarm')
plt.title('Features Correlation Heatmap')
plt.show()
Train test split + data standardization + shuffling¶
- Using min max scaler to standardize the data, as features were quite different distributed
- Test size = 0.25, as we do not have too much data
In [67]:
X_train, X_test, y_train, y_test = train_test_split(features, targets, test_size=0.25, random_state=36)
scaler = MinMaxScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)
X_train_shuffled, y_train_shuffled = shuffle(X_train, y_train, random_state=36)
Building model with Keras¶
- Input shape = 30, output shape = probability (value between 0 and 1)
- 7 layers, 8 neurons in the hidden layers, he_normal weight initializer, l2 + dropout regularization, final sigmoid activation
In [69]:
model_1 = keras.models.Sequential([ # more layers and regularization
keras.layers.Flatten(input_shape=[30]),
keras.layers.Dense(8, activation='relu', kernel_initializer = 'he_normal'),
keras.layers.Dense(8, activation='relu', kernel_regularizer = keras.regularizers.l2(0.01)),
keras.layers.Dropout(rate=0.1),
keras.layers.Dense(8, activation='relu'), #kernel_regularizer = keras.regularizers.l2(0.01)),
keras.layers.Dropout(rate=0.1),
keras.layers.Dense(1, activation='sigmoid')
])
In [70]:
print(model_1.summary())
# As we are using dropout during training, there is a higher loss during training than during validation! The model is hence more robust during val than train due to dropout
Model: "sequential_2"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
flatten_2 (Flatten) (None, 30) 0
dense_8 (Dense) (None, 8) 248
dense_9 (Dense) (None, 8) 72
dropout_4 (Dropout) (None, 8) 0
dense_10 (Dense) (None, 8) 72
dropout_5 (Dropout) (None, 8) 0
dense_11 (Dense) (None, 1) 9
=================================================================
Total params: 401 (1.57 KB)
Trainable params: 401 (1.57 KB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________
None
Training the model¶
- Loss = Binary Cross Entropy
- Optimizer = Adam at 0.001 learning rate
- Metric = accuracy
- Batch size = 64
- Epochs = 300
In [71]:
model_1.compile(loss='BinaryCrossentropy', optimizer=keras.optimizers.Adam(learning_rate=0.001), metrics = 'accuracy')
history = model_1.fit(X_train_shuffled, y_train_shuffled, batch_size = 64, epochs = 300, validation_data = (X_test, y_test))
Epoch 1/300 7/7 [==============================] - 1s 27ms/step - loss: 0.7831 - accuracy: 0.5493 - val_loss: 0.7472 - val_accuracy: 0.6084 Epoch 2/300 7/7 [==============================] - 0s 6ms/step - loss: 0.7538 - accuracy: 0.6150 - val_loss: 0.7297 - val_accuracy: 0.6573 Epoch 3/300 7/7 [==============================] - 0s 7ms/step - loss: 0.7436 - accuracy: 0.6362 - val_loss: 0.7167 - val_accuracy: 0.6783 Epoch 4/300 7/7 [==============================] - 0s 5ms/step - loss: 0.7284 - accuracy: 0.6714 - val_loss: 0.7039 - val_accuracy: 0.6923 Epoch 5/300 7/7 [==============================] - 0s 5ms/step - loss: 0.7156 - accuracy: 0.6925 - val_loss: 0.6909 - val_accuracy: 0.7483 Epoch 6/300 7/7 [==============================] - 0s 5ms/step - loss: 0.7008 - accuracy: 0.7230 - val_loss: 0.6788 - val_accuracy: 0.7692 Epoch 7/300 7/7 [==============================] - 0s 5ms/step - loss: 0.6944 - accuracy: 0.7371 - val_loss: 0.6665 - val_accuracy: 0.7902 Epoch 8/300 7/7 [==============================] - 0s 5ms/step - loss: 0.6834 - accuracy: 0.7746 - val_loss: 0.6537 - val_accuracy: 0.8112 Epoch 9/300 7/7 [==============================] - 0s 8ms/step - loss: 0.6603 - accuracy: 0.7911 - val_loss: 0.6404 - val_accuracy: 0.8182 Epoch 10/300 7/7 [==============================] - 0s 8ms/step - loss: 0.6483 - accuracy: 0.7723 - val_loss: 0.6267 - val_accuracy: 0.8182 Epoch 11/300 7/7 [==============================] - 0s 8ms/step - loss: 0.6429 - accuracy: 0.7840 - val_loss: 0.6125 - val_accuracy: 0.8252 Epoch 12/300 7/7 [==============================] - 0s 8ms/step - loss: 0.6259 - accuracy: 0.7911 - val_loss: 0.5978 - val_accuracy: 0.8392 Epoch 13/300 7/7 [==============================] - 0s 5ms/step - loss: 0.6103 - accuracy: 0.8099 - val_loss: 0.5823 - val_accuracy: 0.8392 Epoch 14/300 7/7 [==============================] - 0s 8ms/step - loss: 0.5911 - accuracy: 0.8263 - val_loss: 0.5665 - val_accuracy: 0.8392 Epoch 15/300 7/7 [==============================] - 0s 5ms/step - loss: 0.5938 - accuracy: 0.7958 - val_loss: 0.5497 - val_accuracy: 0.8671 Epoch 16/300 7/7 [==============================] - 0s 5ms/step - loss: 0.5699 - accuracy: 0.8239 - val_loss: 0.5333 - val_accuracy: 0.8601 Epoch 17/300 7/7 [==============================] - 0s 5ms/step - loss: 0.5559 - accuracy: 0.8310 - val_loss: 0.5168 - val_accuracy: 0.8671 Epoch 18/300 7/7 [==============================] - 0s 5ms/step - loss: 0.5446 - accuracy: 0.8333 - val_loss: 0.4994 - val_accuracy: 0.9021 Epoch 19/300 7/7 [==============================] - 0s 7ms/step - loss: 0.5204 - accuracy: 0.8380 - val_loss: 0.4823 - val_accuracy: 0.9091 Epoch 20/300 7/7 [==============================] - 0s 8ms/step - loss: 0.5075 - accuracy: 0.8427 - val_loss: 0.4654 - val_accuracy: 0.9231 Epoch 21/300 7/7 [==============================] - 0s 11ms/step - loss: 0.5134 - accuracy: 0.8286 - val_loss: 0.4486 - val_accuracy: 0.9510 Epoch 22/300 7/7 [==============================] - 0s 8ms/step - loss: 0.4739 - accuracy: 0.8615 - val_loss: 0.4313 - val_accuracy: 0.9510 Epoch 23/300 7/7 [==============================] - 0s 5ms/step - loss: 0.4628 - accuracy: 0.8685 - val_loss: 0.4150 - val_accuracy: 0.9510 Epoch 24/300 7/7 [==============================] - 0s 8ms/step - loss: 0.4523 - accuracy: 0.8545 - val_loss: 0.3992 - val_accuracy: 0.9510 Epoch 25/300 7/7 [==============================] - 0s 7ms/step - loss: 0.4370 - accuracy: 0.8803 - val_loss: 0.3837 - val_accuracy: 0.9510 Epoch 26/300 7/7 [==============================] - 0s 5ms/step - loss: 0.4254 - accuracy: 0.8756 - val_loss: 0.3687 - val_accuracy: 0.9510 Epoch 27/300 7/7 [==============================] - 0s 8ms/step - loss: 0.4141 - accuracy: 0.8826 - val_loss: 0.3543 - val_accuracy: 0.9510 Epoch 28/300 7/7 [==============================] - 0s 5ms/step - loss: 0.4085 - accuracy: 0.8662 - val_loss: 0.3416 - val_accuracy: 0.9510 Epoch 29/300 7/7 [==============================] - 0s 10ms/step - loss: 0.3941 - accuracy: 0.8709 - val_loss: 0.3285 - val_accuracy: 0.9510 Epoch 30/300 7/7 [==============================] - 0s 7ms/step - loss: 0.3657 - accuracy: 0.8944 - val_loss: 0.3172 - val_accuracy: 0.9510 Epoch 31/300 7/7 [==============================] - 0s 8ms/step - loss: 0.3679 - accuracy: 0.8756 - val_loss: 0.3063 - val_accuracy: 0.9510 Epoch 32/300 7/7 [==============================] - 0s 7ms/step - loss: 0.3647 - accuracy: 0.8803 - val_loss: 0.2964 - val_accuracy: 0.9510 Epoch 33/300 7/7 [==============================] - 0s 8ms/step - loss: 0.3444 - accuracy: 0.8991 - val_loss: 0.2873 - val_accuracy: 0.9510 Epoch 34/300 7/7 [==============================] - 0s 7ms/step - loss: 0.3479 - accuracy: 0.8897 - val_loss: 0.2784 - val_accuracy: 0.9510 Epoch 35/300 7/7 [==============================] - 0s 7ms/step - loss: 0.3364 - accuracy: 0.8873 - val_loss: 0.2705 - val_accuracy: 0.9510 Epoch 36/300 7/7 [==============================] - 0s 6ms/step - loss: 0.3532 - accuracy: 0.8779 - val_loss: 0.2649 - val_accuracy: 0.9580 Epoch 37/300 7/7 [==============================] - 0s 6ms/step - loss: 0.3224 - accuracy: 0.8920 - val_loss: 0.2564 - val_accuracy: 0.9441 Epoch 38/300 7/7 [==============================] - 0s 8ms/step - loss: 0.3257 - accuracy: 0.8897 - val_loss: 0.2515 - val_accuracy: 0.9510 Epoch 39/300 7/7 [==============================] - 0s 7ms/step - loss: 0.3111 - accuracy: 0.8873 - val_loss: 0.2442 - val_accuracy: 0.9510 Epoch 40/300 7/7 [==============================] - 0s 10ms/step - loss: 0.3218 - accuracy: 0.8897 - val_loss: 0.2414 - val_accuracy: 0.9720 Epoch 41/300 7/7 [==============================] - 0s 8ms/step - loss: 0.2949 - accuracy: 0.9061 - val_loss: 0.2326 - val_accuracy: 0.9441 Epoch 42/300 7/7 [==============================] - 0s 8ms/step - loss: 0.3021 - accuracy: 0.8944 - val_loss: 0.2276 - val_accuracy: 0.9510 Epoch 43/300 7/7 [==============================] - 0s 4ms/step - loss: 0.2910 - accuracy: 0.9014 - val_loss: 0.2227 - val_accuracy: 0.9510 Epoch 44/300 7/7 [==============================] - 0s 6ms/step - loss: 0.2732 - accuracy: 0.9155 - val_loss: 0.2172 - val_accuracy: 0.9510 Epoch 45/300 7/7 [==============================] - 0s 5ms/step - loss: 0.2841 - accuracy: 0.9014 - val_loss: 0.2113 - val_accuracy: 0.9510 Epoch 46/300 7/7 [==============================] - 0s 5ms/step - loss: 0.2802 - accuracy: 0.9085 - val_loss: 0.2064 - val_accuracy: 0.9580 Epoch 47/300 7/7 [==============================] - 0s 5ms/step - loss: 0.2654 - accuracy: 0.9296 - val_loss: 0.2002 - val_accuracy: 0.9510 Epoch 48/300 7/7 [==============================] - 0s 5ms/step - loss: 0.2359 - accuracy: 0.9507 - val_loss: 0.1947 - val_accuracy: 0.9510 Epoch 49/300 7/7 [==============================] - 0s 7ms/step - loss: 0.2643 - accuracy: 0.9319 - val_loss: 0.1897 - val_accuracy: 0.9510 Epoch 50/300 7/7 [==============================] - 0s 14ms/step - loss: 0.2561 - accuracy: 0.9484 - val_loss: 0.1847 - val_accuracy: 0.9580 Epoch 51/300 7/7 [==============================] - 0s 5ms/step - loss: 0.2469 - accuracy: 0.9413 - val_loss: 0.1795 - val_accuracy: 0.9580 Epoch 52/300 7/7 [==============================] - 0s 5ms/step - loss: 0.2457 - accuracy: 0.9319 - val_loss: 0.1749 - val_accuracy: 0.9580 Epoch 53/300 7/7 [==============================] - 0s 10ms/step - loss: 0.2352 - accuracy: 0.9484 - val_loss: 0.1688 - val_accuracy: 0.9650 Epoch 54/300 7/7 [==============================] - 0s 9ms/step - loss: 0.2281 - accuracy: 0.9531 - val_loss: 0.1640 - val_accuracy: 0.9650 Epoch 55/300 7/7 [==============================] - 0s 8ms/step - loss: 0.2214 - accuracy: 0.9366 - val_loss: 0.1601 - val_accuracy: 0.9580 Epoch 56/300 7/7 [==============================] - 0s 5ms/step - loss: 0.2102 - accuracy: 0.9460 - val_loss: 0.1561 - val_accuracy: 0.9650 Epoch 57/300 7/7 [==============================] - 0s 8ms/step - loss: 0.2290 - accuracy: 0.9437 - val_loss: 0.1524 - val_accuracy: 0.9650 Epoch 58/300 7/7 [==============================] - 0s 8ms/step - loss: 0.2064 - accuracy: 0.9460 - val_loss: 0.1505 - val_accuracy: 0.9650 Epoch 59/300 7/7 [==============================] - 0s 8ms/step - loss: 0.2277 - accuracy: 0.9319 - val_loss: 0.1476 - val_accuracy: 0.9720 Epoch 60/300 7/7 [==============================] - 0s 8ms/step - loss: 0.2055 - accuracy: 0.9484 - val_loss: 0.1453 - val_accuracy: 0.9650 Epoch 61/300 7/7 [==============================] - 0s 8ms/step - loss: 0.2120 - accuracy: 0.9390 - val_loss: 0.1438 - val_accuracy: 0.9790 Epoch 62/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1981 - accuracy: 0.9531 - val_loss: 0.1410 - val_accuracy: 0.9720 Epoch 63/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1874 - accuracy: 0.9577 - val_loss: 0.1394 - val_accuracy: 0.9720 Epoch 64/300 7/7 [==============================] - 0s 7ms/step - loss: 0.2003 - accuracy: 0.9531 - val_loss: 0.1371 - val_accuracy: 0.9720 Epoch 65/300 7/7 [==============================] - 0s 11ms/step - loss: 0.1883 - accuracy: 0.9343 - val_loss: 0.1344 - val_accuracy: 0.9790 Epoch 66/300 7/7 [==============================] - 0s 10ms/step - loss: 0.1805 - accuracy: 0.9577 - val_loss: 0.1329 - val_accuracy: 0.9790 Epoch 67/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1770 - accuracy: 0.9507 - val_loss: 0.1310 - val_accuracy: 0.9860 Epoch 68/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1812 - accuracy: 0.9531 - val_loss: 0.1301 - val_accuracy: 0.9720 Epoch 69/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1757 - accuracy: 0.9601 - val_loss: 0.1275 - val_accuracy: 0.9790 Epoch 70/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1928 - accuracy: 0.9437 - val_loss: 0.1259 - val_accuracy: 0.9790 Epoch 71/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1673 - accuracy: 0.9624 - val_loss: 0.1246 - val_accuracy: 0.9860 Epoch 72/300 7/7 [==============================] - 0s 9ms/step - loss: 0.1673 - accuracy: 0.9577 - val_loss: 0.1245 - val_accuracy: 0.9790 Epoch 73/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1766 - accuracy: 0.9554 - val_loss: 0.1218 - val_accuracy: 0.9860 Epoch 74/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1754 - accuracy: 0.9484 - val_loss: 0.1216 - val_accuracy: 0.9790 Epoch 75/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1671 - accuracy: 0.9531 - val_loss: 0.1201 - val_accuracy: 0.9860 Epoch 76/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1679 - accuracy: 0.9507 - val_loss: 0.1237 - val_accuracy: 0.9720 Epoch 77/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1651 - accuracy: 0.9601 - val_loss: 0.1177 - val_accuracy: 0.9790 Epoch 78/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1712 - accuracy: 0.9554 - val_loss: 0.1172 - val_accuracy: 0.9790 Epoch 79/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1546 - accuracy: 0.9554 - val_loss: 0.1155 - val_accuracy: 0.9860 Epoch 80/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1524 - accuracy: 0.9577 - val_loss: 0.1139 - val_accuracy: 0.9860 Epoch 81/300 7/7 [==============================] - 0s 14ms/step - loss: 0.1532 - accuracy: 0.9601 - val_loss: 0.1132 - val_accuracy: 0.9860 Epoch 82/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1533 - accuracy: 0.9671 - val_loss: 0.1116 - val_accuracy: 0.9860 Epoch 83/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1602 - accuracy: 0.9624 - val_loss: 0.1107 - val_accuracy: 0.9790 Epoch 84/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1470 - accuracy: 0.9601 - val_loss: 0.1103 - val_accuracy: 0.9790 Epoch 85/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1485 - accuracy: 0.9648 - val_loss: 0.1090 - val_accuracy: 0.9860 Epoch 86/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1418 - accuracy: 0.9695 - val_loss: 0.1084 - val_accuracy: 0.9860 Epoch 87/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1392 - accuracy: 0.9718 - val_loss: 0.1086 - val_accuracy: 0.9860 Epoch 88/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1342 - accuracy: 0.9718 - val_loss: 0.1063 - val_accuracy: 0.9860 Epoch 89/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1533 - accuracy: 0.9671 - val_loss: 0.1056 - val_accuracy: 0.9860 Epoch 90/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1501 - accuracy: 0.9624 - val_loss: 0.1050 - val_accuracy: 0.9860 Epoch 91/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1448 - accuracy: 0.9624 - val_loss: 0.1043 - val_accuracy: 0.9860 Epoch 92/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1362 - accuracy: 0.9624 - val_loss: 0.1038 - val_accuracy: 0.9860 Epoch 93/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1307 - accuracy: 0.9765 - val_loss: 0.1031 - val_accuracy: 0.9860 Epoch 94/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1358 - accuracy: 0.9765 - val_loss: 0.1026 - val_accuracy: 0.9860 Epoch 95/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1298 - accuracy: 0.9695 - val_loss: 0.1017 - val_accuracy: 0.9860 Epoch 96/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1362 - accuracy: 0.9648 - val_loss: 0.1011 - val_accuracy: 0.9860 Epoch 97/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1462 - accuracy: 0.9577 - val_loss: 0.1011 - val_accuracy: 0.9860 Epoch 98/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1526 - accuracy: 0.9507 - val_loss: 0.1014 - val_accuracy: 0.9860 Epoch 99/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1428 - accuracy: 0.9577 - val_loss: 0.1000 - val_accuracy: 0.9860 Epoch 100/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1276 - accuracy: 0.9718 - val_loss: 0.1001 - val_accuracy: 0.9860 Epoch 101/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1342 - accuracy: 0.9695 - val_loss: 0.0989 - val_accuracy: 0.9860 Epoch 102/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1366 - accuracy: 0.9648 - val_loss: 0.0997 - val_accuracy: 0.9860 Epoch 103/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1369 - accuracy: 0.9671 - val_loss: 0.0980 - val_accuracy: 0.9860 Epoch 104/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1347 - accuracy: 0.9671 - val_loss: 0.0981 - val_accuracy: 0.9860 Epoch 105/300 7/7 [==============================] - 0s 14ms/step - loss: 0.1288 - accuracy: 0.9648 - val_loss: 0.0974 - val_accuracy: 0.9860 Epoch 106/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1194 - accuracy: 0.9718 - val_loss: 0.0964 - val_accuracy: 0.9860 Epoch 107/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1234 - accuracy: 0.9789 - val_loss: 0.0962 - val_accuracy: 0.9860 Epoch 108/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1285 - accuracy: 0.9695 - val_loss: 0.0954 - val_accuracy: 0.9860 Epoch 109/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1319 - accuracy: 0.9624 - val_loss: 0.0951 - val_accuracy: 0.9930 Epoch 110/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1164 - accuracy: 0.9718 - val_loss: 0.0953 - val_accuracy: 0.9860 Epoch 111/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1226 - accuracy: 0.9718 - val_loss: 0.0937 - val_accuracy: 0.9860 Epoch 112/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1241 - accuracy: 0.9718 - val_loss: 0.0932 - val_accuracy: 0.9860 Epoch 113/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1345 - accuracy: 0.9624 - val_loss: 0.0926 - val_accuracy: 0.9860 Epoch 114/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1313 - accuracy: 0.9718 - val_loss: 0.0924 - val_accuracy: 0.9860 Epoch 115/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1205 - accuracy: 0.9742 - val_loss: 0.0921 - val_accuracy: 0.9860 Epoch 116/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1349 - accuracy: 0.9577 - val_loss: 0.0920 - val_accuracy: 0.9930 Epoch 117/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1120 - accuracy: 0.9765 - val_loss: 0.0915 - val_accuracy: 0.9860 Epoch 118/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1283 - accuracy: 0.9648 - val_loss: 0.0912 - val_accuracy: 0.9860 Epoch 119/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1177 - accuracy: 0.9742 - val_loss: 0.0909 - val_accuracy: 0.9860 Epoch 120/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1303 - accuracy: 0.9648 - val_loss: 0.0908 - val_accuracy: 0.9930 Epoch 121/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1273 - accuracy: 0.9718 - val_loss: 0.0907 - val_accuracy: 0.9860 Epoch 122/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1219 - accuracy: 0.9765 - val_loss: 0.0911 - val_accuracy: 0.9860 Epoch 123/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1252 - accuracy: 0.9695 - val_loss: 0.0904 - val_accuracy: 0.9860 Epoch 124/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1187 - accuracy: 0.9695 - val_loss: 0.0901 - val_accuracy: 0.9930 Epoch 125/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1337 - accuracy: 0.9624 - val_loss: 0.0922 - val_accuracy: 0.9860 Epoch 126/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1182 - accuracy: 0.9718 - val_loss: 0.0894 - val_accuracy: 0.9860 Epoch 127/300 7/7 [==============================] - 0s 13ms/step - loss: 0.1162 - accuracy: 0.9695 - val_loss: 0.0889 - val_accuracy: 0.9860 Epoch 128/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1141 - accuracy: 0.9695 - val_loss: 0.0887 - val_accuracy: 0.9930 Epoch 129/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1076 - accuracy: 0.9765 - val_loss: 0.0887 - val_accuracy: 0.9860 Epoch 130/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1173 - accuracy: 0.9601 - val_loss: 0.0876 - val_accuracy: 0.9860 Epoch 131/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1161 - accuracy: 0.9742 - val_loss: 0.0872 - val_accuracy: 0.9860 Epoch 132/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1095 - accuracy: 0.9789 - val_loss: 0.0876 - val_accuracy: 0.9860 Epoch 133/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1091 - accuracy: 0.9765 - val_loss: 0.0871 - val_accuracy: 0.9930 Epoch 134/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1074 - accuracy: 0.9765 - val_loss: 0.0863 - val_accuracy: 0.9860 Epoch 135/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1069 - accuracy: 0.9765 - val_loss: 0.0858 - val_accuracy: 0.9930 Epoch 136/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1166 - accuracy: 0.9695 - val_loss: 0.0859 - val_accuracy: 0.9930 Epoch 137/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1131 - accuracy: 0.9765 - val_loss: 0.0855 - val_accuracy: 0.9860 Epoch 138/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1132 - accuracy: 0.9742 - val_loss: 0.0852 - val_accuracy: 0.9930 Epoch 139/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1047 - accuracy: 0.9718 - val_loss: 0.0850 - val_accuracy: 0.9860 Epoch 140/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1047 - accuracy: 0.9789 - val_loss: 0.0847 - val_accuracy: 0.9930 Epoch 141/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1091 - accuracy: 0.9742 - val_loss: 0.0844 - val_accuracy: 0.9930 Epoch 142/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1055 - accuracy: 0.9765 - val_loss: 0.0840 - val_accuracy: 0.9930 Epoch 143/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1003 - accuracy: 0.9718 - val_loss: 0.0839 - val_accuracy: 0.9860 Epoch 144/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1015 - accuracy: 0.9789 - val_loss: 0.0835 - val_accuracy: 0.9860 Epoch 145/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1025 - accuracy: 0.9765 - val_loss: 0.0831 - val_accuracy: 0.9860 Epoch 146/300 7/7 [==============================] - 0s 13ms/step - loss: 0.1071 - accuracy: 0.9765 - val_loss: 0.0831 - val_accuracy: 0.9930 Epoch 147/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0941 - accuracy: 0.9812 - val_loss: 0.0826 - val_accuracy: 0.9860 Epoch 148/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1040 - accuracy: 0.9718 - val_loss: 0.0823 - val_accuracy: 0.9930 Epoch 149/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1050 - accuracy: 0.9765 - val_loss: 0.0821 - val_accuracy: 0.9860 Epoch 150/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0975 - accuracy: 0.9789 - val_loss: 0.0819 - val_accuracy: 0.9930 Epoch 151/300 7/7 [==============================] - 0s 7ms/step - loss: 0.1030 - accuracy: 0.9789 - val_loss: 0.0820 - val_accuracy: 0.9860 Epoch 152/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0989 - accuracy: 0.9789 - val_loss: 0.0816 - val_accuracy: 0.9930 Epoch 153/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1119 - accuracy: 0.9671 - val_loss: 0.0810 - val_accuracy: 0.9930 Epoch 154/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1087 - accuracy: 0.9718 - val_loss: 0.0808 - val_accuracy: 0.9930 Epoch 155/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0973 - accuracy: 0.9789 - val_loss: 0.0823 - val_accuracy: 0.9860 Epoch 156/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1063 - accuracy: 0.9765 - val_loss: 0.0809 - val_accuracy: 0.9860 Epoch 157/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1009 - accuracy: 0.9742 - val_loss: 0.0801 - val_accuracy: 0.9930 Epoch 158/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1017 - accuracy: 0.9742 - val_loss: 0.0802 - val_accuracy: 0.9860 Epoch 159/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0980 - accuracy: 0.9812 - val_loss: 0.0797 - val_accuracy: 0.9860 Epoch 160/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1055 - accuracy: 0.9742 - val_loss: 0.0799 - val_accuracy: 0.9860 Epoch 161/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0979 - accuracy: 0.9718 - val_loss: 0.0796 - val_accuracy: 0.9860 Epoch 162/300 7/7 [==============================] - 0s 13ms/step - loss: 0.0962 - accuracy: 0.9789 - val_loss: 0.0794 - val_accuracy: 0.9860 Epoch 163/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1011 - accuracy: 0.9742 - val_loss: 0.0794 - val_accuracy: 0.9860 Epoch 164/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0918 - accuracy: 0.9765 - val_loss: 0.0782 - val_accuracy: 0.9930 Epoch 165/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0972 - accuracy: 0.9718 - val_loss: 0.0785 - val_accuracy: 0.9860 Epoch 166/300 7/7 [==============================] - 0s 10ms/step - loss: 0.1014 - accuracy: 0.9718 - val_loss: 0.0781 - val_accuracy: 0.9860 Epoch 167/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0898 - accuracy: 0.9765 - val_loss: 0.0775 - val_accuracy: 0.9930 Epoch 168/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0962 - accuracy: 0.9742 - val_loss: 0.0772 - val_accuracy: 0.9930 Epoch 169/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1023 - accuracy: 0.9718 - val_loss: 0.0770 - val_accuracy: 0.9930 Epoch 170/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1067 - accuracy: 0.9742 - val_loss: 0.0770 - val_accuracy: 0.9930 Epoch 171/300 7/7 [==============================] - 0s 5ms/step - loss: 0.1093 - accuracy: 0.9742 - val_loss: 0.0776 - val_accuracy: 0.9860 Epoch 172/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0994 - accuracy: 0.9789 - val_loss: 0.0771 - val_accuracy: 0.9930 Epoch 173/300 7/7 [==============================] - 0s 8ms/step - loss: 0.1057 - accuracy: 0.9671 - val_loss: 0.0789 - val_accuracy: 0.9860 Epoch 174/300 7/7 [==============================] - 0s 9ms/step - loss: 0.0984 - accuracy: 0.9812 - val_loss: 0.0795 - val_accuracy: 0.9860 Epoch 175/300 7/7 [==============================] - 0s 9ms/step - loss: 0.1009 - accuracy: 0.9742 - val_loss: 0.0806 - val_accuracy: 0.9790 Epoch 176/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0995 - accuracy: 0.9718 - val_loss: 0.0784 - val_accuracy: 0.9860 Epoch 177/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0965 - accuracy: 0.9695 - val_loss: 0.0785 - val_accuracy: 0.9860 Epoch 178/300 7/7 [==============================] - 0s 13ms/step - loss: 0.1003 - accuracy: 0.9671 - val_loss: 0.0761 - val_accuracy: 0.9930 Epoch 179/300 7/7 [==============================] - 0s 6ms/step - loss: 0.1125 - accuracy: 0.9695 - val_loss: 0.0843 - val_accuracy: 0.9790 Epoch 180/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0987 - accuracy: 0.9601 - val_loss: 0.0763 - val_accuracy: 0.9930 Epoch 181/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0823 - accuracy: 0.9789 - val_loss: 0.0770 - val_accuracy: 0.9860 Epoch 182/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0927 - accuracy: 0.9789 - val_loss: 0.0757 - val_accuracy: 0.9930 Epoch 183/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0928 - accuracy: 0.9789 - val_loss: 0.0758 - val_accuracy: 0.9930 Epoch 184/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0992 - accuracy: 0.9789 - val_loss: 0.0752 - val_accuracy: 0.9930 Epoch 185/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0918 - accuracy: 0.9836 - val_loss: 0.0751 - val_accuracy: 0.9930 Epoch 186/300 7/7 [==============================] - 0s 9ms/step - loss: 0.0940 - accuracy: 0.9742 - val_loss: 0.0752 - val_accuracy: 0.9930 Epoch 187/300 7/7 [==============================] - 0s 9ms/step - loss: 0.0862 - accuracy: 0.9765 - val_loss: 0.0749 - val_accuracy: 0.9930 Epoch 188/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0892 - accuracy: 0.9812 - val_loss: 0.0751 - val_accuracy: 0.9860 Epoch 189/300 7/7 [==============================] - 0s 11ms/step - loss: 0.0906 - accuracy: 0.9765 - val_loss: 0.0746 - val_accuracy: 0.9930 Epoch 190/300 7/7 [==============================] - 0s 9ms/step - loss: 0.0841 - accuracy: 0.9836 - val_loss: 0.0742 - val_accuracy: 0.9930 Epoch 191/300 7/7 [==============================] - 0s 12ms/step - loss: 0.0938 - accuracy: 0.9742 - val_loss: 0.0745 - val_accuracy: 0.9860 Epoch 192/300 7/7 [==============================] - 0s 11ms/step - loss: 0.0940 - accuracy: 0.9789 - val_loss: 0.0746 - val_accuracy: 0.9860 Epoch 193/300 7/7 [==============================] - 0s 10ms/step - loss: 0.0890 - accuracy: 0.9765 - val_loss: 0.0746 - val_accuracy: 0.9860 Epoch 194/300 7/7 [==============================] - 0s 14ms/step - loss: 0.0912 - accuracy: 0.9765 - val_loss: 0.0740 - val_accuracy: 0.9930 Epoch 195/300 7/7 [==============================] - 0s 10ms/step - loss: 0.0842 - accuracy: 0.9789 - val_loss: 0.0739 - val_accuracy: 0.9860 Epoch 196/300 7/7 [==============================] - 0s 12ms/step - loss: 0.1003 - accuracy: 0.9695 - val_loss: 0.0735 - val_accuracy: 0.9930 Epoch 197/300 7/7 [==============================] - 0s 13ms/step - loss: 0.0971 - accuracy: 0.9718 - val_loss: 0.0741 - val_accuracy: 0.9860 Epoch 198/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0976 - accuracy: 0.9765 - val_loss: 0.0739 - val_accuracy: 0.9930 Epoch 199/300 7/7 [==============================] - 0s 9ms/step - loss: 0.0911 - accuracy: 0.9742 - val_loss: 0.0733 - val_accuracy: 0.9930 Epoch 200/300 7/7 [==============================] - 0s 9ms/step - loss: 0.0958 - accuracy: 0.9695 - val_loss: 0.0787 - val_accuracy: 0.9790 Epoch 201/300 7/7 [==============================] - 0s 9ms/step - loss: 0.0908 - accuracy: 0.9765 - val_loss: 0.0737 - val_accuracy: 0.9860 Epoch 202/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0977 - accuracy: 0.9742 - val_loss: 0.0733 - val_accuracy: 0.9860 Epoch 203/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0800 - accuracy: 0.9836 - val_loss: 0.0728 - val_accuracy: 0.9930 Epoch 204/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0882 - accuracy: 0.9789 - val_loss: 0.0727 - val_accuracy: 0.9930 Epoch 205/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0898 - accuracy: 0.9812 - val_loss: 0.0733 - val_accuracy: 0.9860 Epoch 206/300 7/7 [==============================] - 0s 13ms/step - loss: 0.0904 - accuracy: 0.9812 - val_loss: 0.0727 - val_accuracy: 0.9930 Epoch 207/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0915 - accuracy: 0.9836 - val_loss: 0.0728 - val_accuracy: 0.9930 Epoch 208/300 7/7 [==============================] - 0s 13ms/step - loss: 0.0923 - accuracy: 0.9812 - val_loss: 0.0725 - val_accuracy: 0.9930 Epoch 209/300 7/7 [==============================] - 0s 13ms/step - loss: 0.0855 - accuracy: 0.9765 - val_loss: 0.0726 - val_accuracy: 0.9930 Epoch 210/300 7/7 [==============================] - 0s 12ms/step - loss: 0.0835 - accuracy: 0.9789 - val_loss: 0.0730 - val_accuracy: 0.9860 Epoch 211/300 7/7 [==============================] - 0s 10ms/step - loss: 0.0956 - accuracy: 0.9765 - val_loss: 0.0730 - val_accuracy: 0.9930 Epoch 212/300 7/7 [==============================] - 0s 11ms/step - loss: 0.0819 - accuracy: 0.9812 - val_loss: 0.0731 - val_accuracy: 0.9860 Epoch 213/300 7/7 [==============================] - 0s 12ms/step - loss: 0.0910 - accuracy: 0.9671 - val_loss: 0.0721 - val_accuracy: 0.9930 Epoch 214/300 7/7 [==============================] - 0s 11ms/step - loss: 0.0818 - accuracy: 0.9836 - val_loss: 0.0722 - val_accuracy: 0.9930 Epoch 215/300 7/7 [==============================] - 0s 12ms/step - loss: 0.0819 - accuracy: 0.9859 - val_loss: 0.0717 - val_accuracy: 0.9930 Epoch 216/300 7/7 [==============================] - 0s 11ms/step - loss: 0.0825 - accuracy: 0.9836 - val_loss: 0.0716 - val_accuracy: 0.9930 Epoch 217/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0907 - accuracy: 0.9836 - val_loss: 0.0714 - val_accuracy: 0.9930 Epoch 218/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0891 - accuracy: 0.9789 - val_loss: 0.0729 - val_accuracy: 0.9790 Epoch 219/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0926 - accuracy: 0.9742 - val_loss: 0.0720 - val_accuracy: 0.9860 Epoch 220/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0894 - accuracy: 0.9765 - val_loss: 0.0720 - val_accuracy: 0.9860 Epoch 221/300 7/7 [==============================] - 0s 13ms/step - loss: 0.0802 - accuracy: 0.9812 - val_loss: 0.0714 - val_accuracy: 0.9930 Epoch 222/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0847 - accuracy: 0.9812 - val_loss: 0.0723 - val_accuracy: 0.9790 Epoch 223/300 7/7 [==============================] - 0s 10ms/step - loss: 0.0922 - accuracy: 0.9789 - val_loss: 0.0714 - val_accuracy: 0.9860 Epoch 224/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0908 - accuracy: 0.9812 - val_loss: 0.0736 - val_accuracy: 0.9860 Epoch 225/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0829 - accuracy: 0.9765 - val_loss: 0.0708 - val_accuracy: 0.9930 Epoch 226/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0886 - accuracy: 0.9812 - val_loss: 0.0710 - val_accuracy: 0.9860 Epoch 227/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0874 - accuracy: 0.9742 - val_loss: 0.0706 - val_accuracy: 0.9930 Epoch 228/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0818 - accuracy: 0.9742 - val_loss: 0.0708 - val_accuracy: 0.9930 Epoch 229/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0890 - accuracy: 0.9765 - val_loss: 0.0708 - val_accuracy: 0.9930 Epoch 230/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0865 - accuracy: 0.9765 - val_loss: 0.0706 - val_accuracy: 0.9930 Epoch 231/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0851 - accuracy: 0.9789 - val_loss: 0.0706 - val_accuracy: 0.9930 Epoch 232/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0917 - accuracy: 0.9695 - val_loss: 0.0705 - val_accuracy: 0.9930 Epoch 233/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0808 - accuracy: 0.9836 - val_loss: 0.0725 - val_accuracy: 0.9790 Epoch 234/300 7/7 [==============================] - 0s 4ms/step - loss: 0.0850 - accuracy: 0.9765 - val_loss: 0.0701 - val_accuracy: 0.9930 Epoch 235/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0759 - accuracy: 0.9836 - val_loss: 0.0707 - val_accuracy: 0.9860 Epoch 236/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0800 - accuracy: 0.9883 - val_loss: 0.0700 - val_accuracy: 0.9860 Epoch 237/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0781 - accuracy: 0.9859 - val_loss: 0.0698 - val_accuracy: 0.9930 Epoch 238/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0830 - accuracy: 0.9812 - val_loss: 0.0696 - val_accuracy: 0.9930 Epoch 239/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0705 - accuracy: 0.9812 - val_loss: 0.0702 - val_accuracy: 0.9790 Epoch 240/300 7/7 [==============================] - 0s 4ms/step - loss: 0.0712 - accuracy: 0.9812 - val_loss: 0.0693 - val_accuracy: 0.9930 Epoch 241/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0831 - accuracy: 0.9812 - val_loss: 0.0694 - val_accuracy: 0.9860 Epoch 242/300 7/7 [==============================] - 0s 9ms/step - loss: 0.0859 - accuracy: 0.9836 - val_loss: 0.0689 - val_accuracy: 0.9930 Epoch 243/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0781 - accuracy: 0.9789 - val_loss: 0.0690 - val_accuracy: 0.9930 Epoch 244/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0859 - accuracy: 0.9765 - val_loss: 0.0686 - val_accuracy: 0.9930 Epoch 245/300 7/7 [==============================] - 0s 9ms/step - loss: 0.0803 - accuracy: 0.9859 - val_loss: 0.0688 - val_accuracy: 0.9930 Epoch 246/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0667 - accuracy: 0.9836 - val_loss: 0.0688 - val_accuracy: 0.9860 Epoch 247/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0720 - accuracy: 0.9836 - val_loss: 0.0682 - val_accuracy: 0.9930 Epoch 248/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0659 - accuracy: 0.9859 - val_loss: 0.0686 - val_accuracy: 0.9860 Epoch 249/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0848 - accuracy: 0.9812 - val_loss: 0.0720 - val_accuracy: 0.9790 Epoch 250/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0794 - accuracy: 0.9812 - val_loss: 0.0682 - val_accuracy: 0.9930 Epoch 251/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0730 - accuracy: 0.9789 - val_loss: 0.0701 - val_accuracy: 0.9860 Epoch 252/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0739 - accuracy: 0.9859 - val_loss: 0.0731 - val_accuracy: 0.9790 Epoch 253/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0814 - accuracy: 0.9789 - val_loss: 0.0689 - val_accuracy: 0.9790 Epoch 254/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0681 - accuracy: 0.9859 - val_loss: 0.0679 - val_accuracy: 0.9930 Epoch 255/300 7/7 [==============================] - 0s 13ms/step - loss: 0.0710 - accuracy: 0.9812 - val_loss: 0.0688 - val_accuracy: 0.9860 Epoch 256/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0794 - accuracy: 0.9789 - val_loss: 0.0679 - val_accuracy: 0.9930 Epoch 257/300 7/7 [==============================] - 0s 10ms/step - loss: 0.0840 - accuracy: 0.9765 - val_loss: 0.0685 - val_accuracy: 0.9860 Epoch 258/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0776 - accuracy: 0.9789 - val_loss: 0.0728 - val_accuracy: 0.9790 Epoch 259/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0828 - accuracy: 0.9789 - val_loss: 0.0688 - val_accuracy: 0.9860 Epoch 260/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0832 - accuracy: 0.9789 - val_loss: 0.0684 - val_accuracy: 0.9860 Epoch 261/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0819 - accuracy: 0.9812 - val_loss: 0.0703 - val_accuracy: 0.9790 Epoch 262/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0883 - accuracy: 0.9742 - val_loss: 0.0681 - val_accuracy: 0.9930 Epoch 263/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0771 - accuracy: 0.9906 - val_loss: 0.0683 - val_accuracy: 0.9930 Epoch 264/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0885 - accuracy: 0.9695 - val_loss: 0.0703 - val_accuracy: 0.9790 Epoch 265/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0706 - accuracy: 0.9812 - val_loss: 0.0685 - val_accuracy: 0.9860 Epoch 266/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0753 - accuracy: 0.9836 - val_loss: 0.0681 - val_accuracy: 0.9930 Epoch 267/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0672 - accuracy: 0.9883 - val_loss: 0.0685 - val_accuracy: 0.9860 Epoch 268/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0676 - accuracy: 0.9859 - val_loss: 0.0684 - val_accuracy: 0.9860 Epoch 269/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0727 - accuracy: 0.9789 - val_loss: 0.0677 - val_accuracy: 0.9930 Epoch 270/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0776 - accuracy: 0.9836 - val_loss: 0.0676 - val_accuracy: 0.9930 Epoch 271/300 7/7 [==============================] - 0s 14ms/step - loss: 0.0719 - accuracy: 0.9906 - val_loss: 0.0693 - val_accuracy: 0.9790 Epoch 272/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0724 - accuracy: 0.9859 - val_loss: 0.0675 - val_accuracy: 0.9930 Epoch 273/300 7/7 [==============================] - 0s 11ms/step - loss: 0.0714 - accuracy: 0.9836 - val_loss: 0.0678 - val_accuracy: 0.9930 Epoch 274/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0761 - accuracy: 0.9812 - val_loss: 0.0681 - val_accuracy: 0.9860 Epoch 275/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0816 - accuracy: 0.9836 - val_loss: 0.0686 - val_accuracy: 0.9790 Epoch 276/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0753 - accuracy: 0.9836 - val_loss: 0.0677 - val_accuracy: 0.9860 Epoch 277/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0792 - accuracy: 0.9789 - val_loss: 0.0671 - val_accuracy: 0.9930 Epoch 278/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0757 - accuracy: 0.9789 - val_loss: 0.0685 - val_accuracy: 0.9790 Epoch 279/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0697 - accuracy: 0.9859 - val_loss: 0.0702 - val_accuracy: 0.9790 Epoch 280/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0787 - accuracy: 0.9765 - val_loss: 0.0671 - val_accuracy: 0.9930 Epoch 281/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0746 - accuracy: 0.9836 - val_loss: 0.0676 - val_accuracy: 0.9860 Epoch 282/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0744 - accuracy: 0.9859 - val_loss: 0.0706 - val_accuracy: 0.9790 Epoch 283/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0710 - accuracy: 0.9883 - val_loss: 0.0680 - val_accuracy: 0.9790 Epoch 284/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0702 - accuracy: 0.9859 - val_loss: 0.0670 - val_accuracy: 0.9860 Epoch 285/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0750 - accuracy: 0.9859 - val_loss: 0.0669 - val_accuracy: 0.9930 Epoch 286/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0724 - accuracy: 0.9812 - val_loss: 0.0681 - val_accuracy: 0.9790 Epoch 287/300 7/7 [==============================] - 0s 12ms/step - loss: 0.0717 - accuracy: 0.9859 - val_loss: 0.0680 - val_accuracy: 0.9790 Epoch 288/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0764 - accuracy: 0.9812 - val_loss: 0.0669 - val_accuracy: 0.9930 Epoch 289/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0769 - accuracy: 0.9812 - val_loss: 0.0669 - val_accuracy: 0.9930 Epoch 290/300 7/7 [==============================] - 0s 6ms/step - loss: 0.0872 - accuracy: 0.9671 - val_loss: 0.0703 - val_accuracy: 0.9790 Epoch 291/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0771 - accuracy: 0.9859 - val_loss: 0.0675 - val_accuracy: 0.9860 Epoch 292/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0739 - accuracy: 0.9859 - val_loss: 0.0669 - val_accuracy: 0.9860 Epoch 293/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0686 - accuracy: 0.9836 - val_loss: 0.0695 - val_accuracy: 0.9790 Epoch 294/300 7/7 [==============================] - 0s 5ms/step - loss: 0.0739 - accuracy: 0.9789 - val_loss: 0.0668 - val_accuracy: 0.9930 Epoch 295/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0739 - accuracy: 0.9836 - val_loss: 0.0675 - val_accuracy: 0.9860 Epoch 296/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0697 - accuracy: 0.9859 - val_loss: 0.0667 - val_accuracy: 0.9930 Epoch 297/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0739 - accuracy: 0.9883 - val_loss: 0.0672 - val_accuracy: 0.9860 Epoch 298/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0746 - accuracy: 0.9789 - val_loss: 0.0688 - val_accuracy: 0.9790 Epoch 299/300 7/7 [==============================] - 0s 7ms/step - loss: 0.0791 - accuracy: 0.9789 - val_loss: 0.0679 - val_accuracy: 0.9860 Epoch 300/300 7/7 [==============================] - 0s 8ms/step - loss: 0.0691 - accuracy: 0.9836 - val_loss: 0.0668 - val_accuracy: 0.9860
Plotting loss and accuracy¶
- We can observe that the model correctly learns to classify between benign (0) and malign (1) after 300 epochs
- The validation loss is lower than the training loss, probably because we are using the dropout regularization technique in the training phase of the model, hence making more robust predictions during validation than training
In [72]:
plt.plot(history.history['loss'], label='Training Loss')
plt.plot(history.history['val_loss'], label='Validation Loss')
plt.title('Model Loss')
plt.xlabel('Epoch')
plt.ylabel('Loss')
plt.legend()
plt.show()
In [73]:
plt.plot(history.history['accuracy'], label='Training Acc')
plt.plot(history.history['val_accuracy'], label='Validation Acc')
plt.title('Model Acc')
plt.xlabel('Epoch')
plt.ylabel('Acc')
plt.legend()
plt.show()
Visualizing some accuracy scores to better understand the model results and convergence¶
In [74]:
history.history['accuracy'][0:15]
Out[74]:
[0.5492957830429077, 0.6150234937667847, 0.6361502408981323, 0.67136150598526, 0.6924882531166077, 0.7230046987533569, 0.737089216709137, 0.7746478915214539, 0.7910798192024231, 0.7723004817962646, 0.7840375304222107, 0.7910798192024231, 0.8098591566085815, 0.8262910842895508, 0.7957746386528015]
In [75]:
history.history['accuracy'][285:299]
Out[75]:
[0.9812206625938416, 0.98591548204422, 0.9812206625938416, 0.9812206625938416, 0.9671361446380615, 0.98591548204422, 0.98591548204422, 0.9835680723190308, 0.9788732528686523, 0.9835680723190308, 0.98591548204422, 0.9882628917694092, 0.9788732528686523, 0.9788732528686523]
Comparing model's predictions with the test data for a visual understanding of the results¶
In [76]:
preds = model_1.predict(X_test)
y_pred_classes = (preds > 0.5).astype(int) # turning the predictions (probabilities) into classes (0 or 1) with a 0.5 threshold
1/5 [=====>........................] - ETA: 0s5/5 [==============================] - 0s 1ms/step
In [77]:
preds[0:10]
Out[77]:
array([[9.9982846e-01],
[9.9974132e-01],
[9.9156398e-01],
[9.9484193e-01],
[9.9979740e-01],
[6.2453331e-08],
[1.3036483e-06],
[9.9852651e-01],
[9.8689264e-01],
[9.9840647e-01]], dtype=float32)
In [78]:
y_pred_classes[0:10]
Out[78]:
array([[1],
[1],
[1],
[1],
[1],
[0],
[0],
[1],
[1],
[1]])
In [79]:
y_test[0:10]
Out[79]:
| target | |
|---|---|
| 63 | 1 |
| 525 | 1 |
| 500 | 1 |
| 292 | 1 |
| 46 | 1 |
| 108 | 0 |
| 323 | 0 |
| 386 | 1 |
| 377 | 1 |
| 467 | 1 |
In [80]:
conf_matrix = confusion_matrix(y_test, y_pred_classes)
In [81]:
plt.figure(figsize=(8, 6))
sns.heatmap(conf_matrix, annot=True, fmt='d', cmap='Reds', xticklabels=[0,1], yticklabels=[0,1])
plt.xlabel('Predicted Label')
plt.ylabel('True Label')
plt.title('Confusion Matrix')
plt.show()
In [ ]: