Maintained by: Chitta Ranjan ([email protected])
This is open source code repository for SGT. Sequence Graph Transform extracts the short- and long-term sequence features and embeds them in a finite-dimensional feature space. Importantly, SGT has low computation and can extract any amount of short- to long-term patterns without any increase in the computation. These properties are proved theoretically and demonstrated on real data in this paper: https://arxiv.org/abs/1608.03533.
If using this code or dataset, please cite the following:
[1] Ranjan, Chitta, Samaneh Ebrahimi, and Kamran Paynabar. "Sequence Graph Transform (SGT): A Feature Extraction Function for Sequence Data Mining." arXiv preprint arXiv:1608.03533 (2016).
@article{ranjan2016sequence, title={Sequence Graph Transform (SGT): A Feature Extraction Function for Sequence Data Mining}, author={Ranjan, Chitta and Ebrahimi, Samaneh and Paynabar, Kamran}, journal={arXiv preprint arXiv:1608.03533}, year={2016} }
You can install sgt directly using a pip command.
$ pip install sgt
import numpy as np
import pandas as pd
from itertools import chain
import warnings
########
from sklearn.preprocessing import LabelEncoder
import tensorflow as tf
from keras.datasets import imdb
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense
from tensorflow.keras.layers import LSTM
from tensorflow.keras.layers import Dropout, Activation, Flatten
from tensorflow.keras.layers import Embedding
from tensorflow.keras.preprocessing import sequence
np.random.seed(7) # fix random seed for reproducibility
from sklearn.model_selection import train_test_split, KFold, StratifiedKFold
import sklearn.metrics
import time
from sklearn.decomposition import PCA
from sklearn.cluster import KMeans
import matplotlib.pyplot as plt
%matplotlib inline
from sgt import Sgttf.__version__'2.0.0'
sgt = Sgt()sequence = np.array(["B","B","A","C","A","C","A","A","B","A"])
alphabets = ["A", "B", "C"]
lengthsensitive = True
kappa = 5sgt.getpositions(sequence = sequence, alphabets = alphabets)[('A', (array([2, 4, 6, 7, 9]),)),
('B', (array([0, 1, 8]),)),
('C', (array([3, 5]),))]
sgt.fit(sequence, alphabets, lengthsensitive, kappa, flatten=False)
<style>
.dataframe thead tr:only-child th {
text-align: right;
}
</style>
.dataframe thead th {
text-align: left;
}
.dataframe tbody tr th {
vertical-align: top;
}
| A | B | C | |
|---|---|---|---|
| A | 0.369361 | 0.442463 | 0.537637 |
| B | 0.414884 | 0.468038 | 0.162774 |
| C | 0.454136 | 0.068693 | 0.214492 |
corpus = [["B","B","A","C","A","C","A","A","B","A"], ["C", "Z", "Z", "Z", "D"]]s = sgt.fit_transform(corpus)
print(s)[[0.90616284 1.31002279 2.6184865 0. 0. 0.86569371
1.23042262 0.52543984 0. 0. 1.37141609 0.28262508
1.35335283 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0.
0. ]
[0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0.
0. 0.09157819 0.92166965 0. 0. 0.
0. 0. 0. 0. 0. 0.92166965
1.45182361]]
sequence_test = [['a', 'b'], ['a', 'b', 'c'], ['e', 'f']]sequence_model_test = Sgt(kappa=10, lengthsensitive=True)result_test = sequence_model_test.fit_transform(corpus=sequence_test)result_testarray([[0. , 0.39428342, 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ],
[0. , 0.41059877, 0.15105085, 0. , 0. ,
0. , 0. , 0.41059877, 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ],
[0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0.39428342,
0. , 0. , 0. , 0. , 0. ]])
sequence_model_test.alphabets['a', 'b', 'c', 'e', 'f']
The data used here is taken from www.uniprot.org. This is a public database for proteins. The data contains the protein sequences and their functions. In the following, we will demonstrate
- clustering of the sequences.
- classification of the sequences with the functions as labels.
protein_data=pd.DataFrame.from_csv('../data/protein_classification.csv')
X=protein_data['Sequence']
def split(word):
return [char for char in word]
sequences = [split(x) for x in X]
print(sequences[0])['M', 'E', 'I', 'E', 'K', 'T', 'N', 'R', 'M', 'N', 'A', 'L', 'F', 'E', 'F', 'Y', 'A', 'A', 'L', 'L', 'T', 'D', 'K', 'Q', 'M', 'N', 'Y', 'I', 'E', 'L', 'Y', 'Y', 'A', 'D', 'D', 'Y', 'S', 'L', 'A', 'E', 'I', 'A', 'E', 'E', 'F', 'G', 'V', 'S', 'R', 'Q', 'A', 'V', 'Y', 'D', 'N', 'I', 'K', 'R', 'T', 'E', 'K', 'I', 'L', 'E', 'D', 'Y', 'E', 'M', 'K', 'L', 'H', 'M', 'Y', 'S', 'D', 'Y', 'I', 'V', 'R', 'S', 'Q', 'I', 'F', 'D', 'Q', 'I', 'L', 'E', 'R', 'Y', 'P', 'K', 'D', 'D', 'F', 'L', 'Q', 'E', 'Q', 'I', 'E', 'I', 'L', 'T', 'S', 'I', 'D', 'N', 'R', 'E']
sgt = Sgt(kappa = 1, lengthsensitive = False)embedding = sgt.fit_transform(corpus=sequences)embedding.shape(2112, 400)
We perform PCA on the sequence embeddings and then do kmeans clustering.
pca = PCA(n_components=2)
pca.fit(embedding)
X=pca.transform(embedding)
print(np.sum(pca.explained_variance_ratio_))
df = pd.DataFrame(data=X, columns=['x1', 'x2'])
df.head()0.6432744907364925
<style>
.dataframe thead tr:only-child th {
text-align: right;
}
</style>
.dataframe thead th {
text-align: left;
}
.dataframe tbody tr th {
vertical-align: top;
}
| x1 | x2 | |
|---|---|---|
| 0 | 0.384913 | -0.269873 |
| 1 | 0.022764 | 0.135995 |
| 2 | 0.177792 | -0.172454 |
| 3 | 0.168074 | -0.147334 |
| 4 | 0.383616 | -0.271163 |
kmeans = KMeans(n_clusters=3, max_iter =300)
kmeans.fit(df)
labels = kmeans.predict(df)
centroids = kmeans.cluster_centers_
fig = plt.figure(figsize=(5, 5))
colmap = {1: 'r', 2: 'g', 3: 'b'}
colors = list(map(lambda x: colmap[x+1], labels))
plt.scatter(df['x1'], df['x2'], color=colors, alpha=0.5, edgecolor=colors)<matplotlib.collections.PathCollection at 0x147c494e0>
We perform PCA on the sequence embeddings and then do kmeans clustering.
y = protein_data['Function [CC]']
encoder = LabelEncoder()
encoder.fit(y)
encoded_y = encoder.transform(y)We will perform a 10-fold cross-validation to measure the performance of the classification model.
kfold = 10
X = pd.DataFrame(embedding)
y = encoded_y
random_state = 1
test_F1 = np.zeros(kfold)
skf = KFold(n_splits = kfold, shuffle = True, random_state = random_state)
k = 0
epochs = 50
batch_size = 128
for train_index, test_index in skf.split(X, y):
X_train, X_test = X.iloc[train_index], X.iloc[test_index]
y_train, y_test = y[train_index], y[test_index]
X_train = X_train.as_matrix(columns = None)
X_test = X_test.as_matrix(columns = None)
model = Sequential()
model.add(Dense(64, input_shape = (X_train.shape[1],)))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(32))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(1))
model.add(Activation('sigmoid'))
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
model.fit(X_train, y_train ,batch_size=batch_size, epochs=epochs, verbose=0)
y_pred = model.predict_proba(X_test).round().astype(int)
y_train_pred = model.predict_proba(X_train).round().astype(int)
test_F1[k] = sklearn.metrics.f1_score(y_test, y_pred)
k+=1
print ('Average f1 score', np.mean(test_F1))Average f1 score 1.0
This data sample is taken from https://www.ll.mit.edu/r-d/datasets/1998-darpa-intrusion-detection-evaluation-dataset. This is a network intrusion data containing audit logs and any attack as a positive label. Since, network intrusion is a rare event, the data is unbalanced. Here we will,
- build a sequence classification model to predict a network intrusion.
Each sequence contains in the data is a series of activity, for example, {login, password}. The alphabets in the input data sequences are already encoded into integers. The original sequences data file is also present in the /data directory.
darpa_data = pd.DataFrame.from_csv('../data/darpa_data.csv')
darpa_data.columnsIndex(['seqlen', 'seq', 'class'], dtype='object')
X = darpa_data['seq']
sequences = [x.split('~') for x in X]y = darpa_data['class']
encoder = LabelEncoder()
encoder.fit(y)
y = encoder.transform(y)In this data, the sequence embeddings should be length-sensitive. The lengths are important here because sequences with similar patterns but different lengths can have different labels. Consider a simple example of two sessions: {login, pswd, login, pswd,...} and {login, pswd,...(repeated several times)..., login, pswd}. While the first session can be a regular user mistyping the password once, the other session is possibly an attack to guess the password. Thus, the sequence lengths are as important as the patterns.
sgt_darpa = Sgt(kappa = 5, lengthsensitive = True)embedding = sgt_darpa.fit_transform(corpus=sequences)pd.DataFrame(embedding).to_csv(path_or_buf='tmp.csv', index=False)
pd.DataFrame(embedding).head()
<style>
.dataframe thead tr:only-child th {
text-align: right;
}
</style>
.dataframe thead th {
text-align: left;
}
.dataframe tbody tr th {
vertical-align: top;
}
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | ... | 2391 | 2392 | 2393 | 2394 | 2395 | 2396 | 2397 | 2398 | 2399 | 2400 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0.069114 | 0.0 | 0.000000e+00 | 0.000000e+00 | 0.0 | 0.000000e+00 | 0.000000 | 0.000000e+00 | 0.000000e+00 | 0.000000e+00 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.000000 | 0.000000e+00 | 0.0 | 0.000000e+00 | 0.000000e+00 |
| 1 | 0.000000 | 0.0 | 4.804190e-09 | 7.041516e-10 | 0.0 | 2.004958e-12 | 0.000132 | 1.046458e-07 | 5.863092e-16 | 7.568986e-23 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.540296 | 5.739230e-32 | 0.0 | 0.000000e+00 | 0.000000e+00 |
| 2 | 0.000000 | 0.0 | 0.000000e+00 | 0.000000e+00 | 0.0 | 0.000000e+00 | 0.000000 | 0.000000e+00 | 0.000000e+00 | 0.000000e+00 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.000000 | 0.000000e+00 | 0.0 | 0.000000e+00 | 0.000000e+00 |
| 3 | 0.785666 | 0.0 | 0.000000e+00 | 0.000000e+00 | 0.0 | 0.000000e+00 | 0.000000 | 1.950089e-03 | 2.239981e-04 | 2.343180e-07 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.528133 | 1.576703e-09 | 0.0 | 2.516644e-29 | 1.484843e-57 |
| 4 | 0.000000 | 0.0 | 0.000000e+00 | 0.000000e+00 | 0.0 | 0.000000e+00 | 0.000000 | 0.000000e+00 | 0.000000e+00 | 0.000000e+00 | ... | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.000000 | 0.000000e+00 | 0.0 | 0.000000e+00 | 0.000000e+00 |
5 rows × 2401 columns
The embeddings are sparse. We, therefore, apply PCA on the embeddings.
from sklearn.decomposition import PCA
pca = PCA(n_components=35)
pca.fit(embedding)
X = pca.transform(embedding)
print(np.sum(pca.explained_variance_ratio_))0.9887812984792304
The PCA transforms of the embeddings are used directly as inputs to an MLP classifier.
kfold = 3
random_state = 11
test_F1 = np.zeros(kfold)
time_k = np.zeros(kfold)
skf = StratifiedKFold(n_splits=kfold, shuffle=True, random_state=random_state)
k = 0
epochs = 300
batch_size = 15
# class_weight = {0 : 1., 1: 1.,} # The weights can be changed and made inversely proportional to the class size to improve the accuracy.
class_weight = {0 : 0.12, 1: 0.88,}
for train_index, test_index in skf.split(X, y):
X_train, X_test = X[train_index], X[test_index]
y_train, y_test = y[train_index], y[test_index]
model = Sequential()
model.add(Dense(128, input_shape=(X_train.shape[1],)))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(1))
model.add(Activation('sigmoid'))
model.summary()
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
start_time = time.time()
model.fit(X_train, y_train ,batch_size=batch_size, epochs=epochs, verbose=1, class_weight=class_weight)
end_time = time.time()
time_k[k] = end_time-start_time
y_pred = model.predict_proba(X_test).round().astype(int)
y_train_pred = model.predict_proba(X_train).round().astype(int)
test_F1[k] = sklearn.metrics.f1_score(y_test, y_pred)
k += 1Model: "sequential_12"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense_30 (Dense) (None, 128) 4608
_________________________________________________________________
activation_30 (Activation) (None, 128) 0
_________________________________________________________________
dropout_20 (Dropout) (None, 128) 0
_________________________________________________________________
dense_31 (Dense) (None, 1) 129
_________________________________________________________________
activation_31 (Activation) (None, 1) 0
=================================================================
Total params: 4,737
Trainable params: 4,737
Non-trainable params: 0
_________________________________________________________________
Train on 73 samples
Epoch 1/300
73/73 [==============================] - 1s 9ms/sample - loss: 0.1489 - accuracy: 0.4658
Epoch 2/300
73/73 [==============================] - 0s 138us/sample - loss: 0.1350 - accuracy: 0.5890
Epoch 3/300
73/73 [==============================] - 0s 138us/sample - loss: 0.1403 - accuracy: 0.5205
Epoch 4/300
73/73 [==============================] - 0s 142us/sample - loss: 0.1272 - accuracy: 0.6849
Epoch 5/300
73/73 [==============================] - 0s 126us/sample - loss: 0.1189 - accuracy: 0.7945
Epoch 6/300
73/73 [==============================] - 0s 131us/sample - loss: 0.1198 - accuracy: 0.7260
Epoch 7/300
73/73 [==============================] - 0s 155us/sample - loss: 0.1100 - accuracy: 0.8904
Epoch 8/300
73/73 [==============================] - 0s 130us/sample - loss: 0.1015 - accuracy: 0.8767
Epoch 9/300
73/73 [==============================] - 0s 146us/sample - loss: 0.0999 - accuracy: 0.8767
Epoch 10/300
73/73 [==============================] - 0s 130us/sample - loss: 0.1011 - accuracy: 0.8356
Epoch 11/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0967 - accuracy: 0.9178
Epoch 12/300
73/73 [==============================] - 0s 140us/sample - loss: 0.0816 - accuracy: 0.9178
Epoch 13/300
73/73 [==============================] - 0s 151us/sample - loss: 0.0858 - accuracy: 0.9041
Epoch 14/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0762 - accuracy: 0.8904
Epoch 15/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0826 - accuracy: 0.8904
Epoch 16/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0757 - accuracy: 0.9178
Epoch 17/300
73/73 [==============================] - 0s 137us/sample - loss: 0.0740 - accuracy: 0.9041
Epoch 18/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0781 - accuracy: 0.9041
Epoch 19/300
73/73 [==============================] - 0s 137us/sample - loss: 0.0696 - accuracy: 0.9178
Epoch 20/300
73/73 [==============================] - 0s 145us/sample - loss: 0.0615 - accuracy: 0.9041
Epoch 21/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0620 - accuracy: 0.9178
Epoch 22/300
73/73 [==============================] - 0s 152us/sample - loss: 0.0618 - accuracy: 0.9041
Epoch 23/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0684 - accuracy: 0.9041
Epoch 24/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0614 - accuracy: 0.9178
Epoch 25/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0594 - accuracy: 0.9041
Epoch 26/300
73/73 [==============================] - 0s 151us/sample - loss: 0.0577 - accuracy: 0.9041
Epoch 27/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0629 - accuracy: 0.9041
Epoch 28/300
73/73 [==============================] - 0s 137us/sample - loss: 0.0488 - accuracy: 0.9178
Epoch 29/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0541 - accuracy: 0.9178
Epoch 30/300
73/73 [==============================] - 0s 142us/sample - loss: 0.0586 - accuracy: 0.9178
Epoch 31/300
73/73 [==============================] - 0s 152us/sample - loss: 0.0521 - accuracy: 0.9041
Epoch 32/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0524 - accuracy: 0.9178
Epoch 33/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0519 - accuracy: 0.9178
Epoch 34/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0490 - accuracy: 0.9178
Epoch 35/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0414 - accuracy: 0.9178
Epoch 36/300
73/73 [==============================] - 0s 155us/sample - loss: 0.0447 - accuracy: 0.9041
Epoch 37/300
73/73 [==============================] - 0s 152us/sample - loss: 0.0413 - accuracy: 0.9178
Epoch 38/300
73/73 [==============================] - 0s 154us/sample - loss: 0.0470 - accuracy: 0.9178
Epoch 39/300
73/73 [==============================] - 0s 161us/sample - loss: 0.0421 - accuracy: 0.9178
Epoch 40/300
73/73 [==============================] - 0s 152us/sample - loss: 0.0431 - accuracy: 0.9178
Epoch 41/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0501 - accuracy: 0.9041
Epoch 42/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0407 - accuracy: 0.9178
Epoch 43/300
73/73 [==============================] - 0s 149us/sample - loss: 0.0389 - accuracy: 0.9178
Epoch 44/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0394 - accuracy: 0.9178
Epoch 45/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0409 - accuracy: 0.9178
Epoch 46/300
73/73 [==============================] - 0s 150us/sample - loss: 0.0403 - accuracy: 0.9178
Epoch 47/300
73/73 [==============================] - 0s 149us/sample - loss: 0.0431 - accuracy: 0.9178
Epoch 48/300
73/73 [==============================] - 0s 158us/sample - loss: 0.0354 - accuracy: 0.9178
Epoch 49/300
73/73 [==============================] - 0s 170us/sample - loss: 0.0420 - accuracy: 0.9178
Epoch 50/300
73/73 [==============================] - 0s 142us/sample - loss: 0.0392 - accuracy: 0.9178
Epoch 51/300
73/73 [==============================] - 0s 167us/sample - loss: 0.0334 - accuracy: 0.9178
Epoch 52/300
73/73 [==============================] - 0s 165us/sample - loss: 0.0352 - accuracy: 0.9178
Epoch 53/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0363 - accuracy: 0.9178
Epoch 54/300
73/73 [==============================] - 0s 150us/sample - loss: 0.0355 - accuracy: 0.9178
Epoch 55/300
73/73 [==============================] - 0s 141us/sample - loss: 0.0373 - accuracy: 0.9178
Epoch 56/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0320 - accuracy: 0.9178
Epoch 57/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0338 - accuracy: 0.9178
Epoch 58/300
73/73 [==============================] - 0s 140us/sample - loss: 0.0332 - accuracy: 0.9178
Epoch 59/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0377 - accuracy: 0.9178
Epoch 60/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0312 - accuracy: 0.9178
Epoch 61/300
73/73 [==============================] - 0s 137us/sample - loss: 0.0344 - accuracy: 0.9178
Epoch 62/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0332 - accuracy: 0.9178
Epoch 63/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0334 - accuracy: 0.9178
Epoch 64/300
73/73 [==============================] - 0s 145us/sample - loss: 0.0347 - accuracy: 0.9178
Epoch 65/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0340 - accuracy: 0.9178
Epoch 66/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0370 - accuracy: 0.9178
Epoch 67/300
73/73 [==============================] - 0s 144us/sample - loss: 0.0335 - accuracy: 0.9178
Epoch 68/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0289 - accuracy: 0.9178
Epoch 69/300
73/73 [==============================] - 0s 124us/sample - loss: 0.0328 - accuracy: 0.9178
Epoch 70/300
73/73 [==============================] - 0s 141us/sample - loss: 0.0350 - accuracy: 0.9178
Epoch 71/300
73/73 [==============================] - 0s 142us/sample - loss: 0.0277 - accuracy: 0.9178
Epoch 72/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0272 - accuracy: 0.9178
Epoch 73/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0292 - accuracy: 0.9178
Epoch 74/300
73/73 [==============================] - 0s 146us/sample - loss: 0.0301 - accuracy: 0.9178
Epoch 75/300
73/73 [==============================] - 0s 141us/sample - loss: 0.0309 - accuracy: 0.9178
Epoch 76/300
73/73 [==============================] - 0s 140us/sample - loss: 0.0269 - accuracy: 0.9178
Epoch 77/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0267 - accuracy: 0.9178
Epoch 78/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0272 - accuracy: 0.9178
Epoch 79/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0318 - accuracy: 0.9178
Epoch 80/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0241 - accuracy: 0.9178
Epoch 81/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0253 - accuracy: 0.9178
Epoch 82/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0248 - accuracy: 0.9178
Epoch 83/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0295 - accuracy: 0.9178
Epoch 84/300
73/73 [==============================] - 0s 127us/sample - loss: 0.0300 - accuracy: 0.9178
Epoch 85/300
73/73 [==============================] - 0s 126us/sample - loss: 0.0220 - accuracy: 0.9315
Epoch 86/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0266 - accuracy: 0.9178
Epoch 87/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0279 - accuracy: 0.9178
Epoch 88/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0318 - accuracy: 0.9178
Epoch 89/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0296 - accuracy: 0.9178
Epoch 90/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0306 - accuracy: 0.9178
Epoch 91/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0234 - accuracy: 0.9178
Epoch 92/300
73/73 [==============================] - 0s 146us/sample - loss: 0.0294 - accuracy: 0.9315
Epoch 93/300
73/73 [==============================] - 0s 125us/sample - loss: 0.0235 - accuracy: 0.9178
Epoch 94/300
73/73 [==============================] - 0s 148us/sample - loss: 0.0305 - accuracy: 0.9178
Epoch 95/300
73/73 [==============================] - 0s 142us/sample - loss: 0.0320 - accuracy: 0.9041
Epoch 96/300
73/73 [==============================] - 0s 124us/sample - loss: 0.0259 - accuracy: 0.9178
Epoch 97/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0264 - accuracy: 0.9178
Epoch 98/300
73/73 [==============================] - 0s 146us/sample - loss: 0.0294 - accuracy: 0.9178
Epoch 99/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0248 - accuracy: 0.9178
Epoch 100/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0256 - accuracy: 0.9178
Epoch 101/300
73/73 [==============================] - 0s 152us/sample - loss: 0.0229 - accuracy: 0.9315
Epoch 102/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0291 - accuracy: 0.9178
Epoch 103/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0224 - accuracy: 0.9178
Epoch 104/300
73/73 [==============================] - 0s 140us/sample - loss: 0.0235 - accuracy: 0.9178
Epoch 105/300
73/73 [==============================] - 0s 147us/sample - loss: 0.0277 - accuracy: 0.9041
Epoch 106/300
73/73 [==============================] - 0s 125us/sample - loss: 0.0219 - accuracy: 0.9178
Epoch 107/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0219 - accuracy: 0.9315
Epoch 108/300
73/73 [==============================] - 0s 140us/sample - loss: 0.0253 - accuracy: 0.9178
Epoch 109/300
73/73 [==============================] - 0s 127us/sample - loss: 0.0243 - accuracy: 0.9315
Epoch 110/300
73/73 [==============================] - 0s 137us/sample - loss: 0.0234 - accuracy: 0.9178
Epoch 111/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0318 - accuracy: 0.9041
Epoch 112/300
73/73 [==============================] - 0s 121us/sample - loss: 0.0215 - accuracy: 0.9178
Epoch 113/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0281 - accuracy: 0.9178
Epoch 114/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0227 - accuracy: 0.9315
Epoch 115/300
73/73 [==============================] - 0s 125us/sample - loss: 0.0270 - accuracy: 0.9178
Epoch 116/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0277 - accuracy: 0.9178
Epoch 117/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0308 - accuracy: 0.9178
Epoch 118/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0287 - accuracy: 0.9315
Epoch 119/300
73/73 [==============================] - 0s 141us/sample - loss: 0.0218 - accuracy: 0.9178
Epoch 120/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0239 - accuracy: 0.9178
Epoch 121/300
73/73 [==============================] - 0s 167us/sample - loss: 0.0254 - accuracy: 0.9178
Epoch 122/300
73/73 [==============================] - 0s 172us/sample - loss: 0.0218 - accuracy: 0.9178
Epoch 123/300
73/73 [==============================] - 0s 174us/sample - loss: 0.0221 - accuracy: 0.9178
Epoch 124/300
73/73 [==============================] - 0s 163us/sample - loss: 0.0272 - accuracy: 0.9178
Epoch 125/300
73/73 [==============================] - 0s 146us/sample - loss: 0.0216 - accuracy: 0.9178
Epoch 126/300
73/73 [==============================] - 0s 148us/sample - loss: 0.0231 - accuracy: 0.9178
Epoch 127/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0228 - accuracy: 0.9178
Epoch 128/300
73/73 [==============================] - 0s 144us/sample - loss: 0.0219 - accuracy: 0.9178
Epoch 129/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0241 - accuracy: 0.9178
Epoch 130/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0245 - accuracy: 0.9178
Epoch 131/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0233 - accuracy: 0.9315
Epoch 132/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0211 - accuracy: 0.9178
Epoch 133/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0221 - accuracy: 0.9178
Epoch 134/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0244 - accuracy: 0.9178
Epoch 135/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0223 - accuracy: 0.9315
Epoch 136/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0332 - accuracy: 0.9041
Epoch 137/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0217 - accuracy: 0.9178
Epoch 138/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0253 - accuracy: 0.9178
Epoch 139/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0203 - accuracy: 0.9178
Epoch 140/300
73/73 [==============================] - 0s 145us/sample - loss: 0.0219 - accuracy: 0.9178
Epoch 141/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0281 - accuracy: 0.9178
Epoch 142/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0206 - accuracy: 0.9178
Epoch 143/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0269 - accuracy: 0.9041
Epoch 144/300
73/73 [==============================] - 0s 144us/sample - loss: 0.0293 - accuracy: 0.9178
Epoch 145/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0290 - accuracy: 0.9178
Epoch 146/300
73/73 [==============================] - 0s 125us/sample - loss: 0.0198 - accuracy: 0.9178
Epoch 147/300
73/73 [==============================] - 0s 137us/sample - loss: 0.0242 - accuracy: 0.9178
Epoch 148/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0218 - accuracy: 0.9178
Epoch 149/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0267 - accuracy: 0.9041
Epoch 150/300
73/73 [==============================] - 0s 149us/sample - loss: 0.0221 - accuracy: 0.9178
Epoch 151/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0222 - accuracy: 0.9178
Epoch 152/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0225 - accuracy: 0.9315
Epoch 153/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0240 - accuracy: 0.9315
Epoch 154/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0218 - accuracy: 0.9178
Epoch 155/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0282 - accuracy: 0.9178
Epoch 156/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0226 - accuracy: 0.9178
Epoch 157/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0232 - accuracy: 0.9178
Epoch 158/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0217 - accuracy: 0.9178
Epoch 159/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0191 - accuracy: 0.9178
Epoch 160/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0214 - accuracy: 0.9178
Epoch 161/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0201 - accuracy: 0.9178
Epoch 162/300
73/73 [==============================] - 0s 126us/sample - loss: 0.0233 - accuracy: 0.9178
Epoch 163/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0217 - accuracy: 0.9178
Epoch 164/300
73/73 [==============================] - 0s 140us/sample - loss: 0.0189 - accuracy: 0.9178
Epoch 165/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0192 - accuracy: 0.9178
Epoch 166/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0230 - accuracy: 0.9178
Epoch 167/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0235 - accuracy: 0.9178
Epoch 168/300
73/73 [==============================] - 0s 142us/sample - loss: 0.0185 - accuracy: 0.9178
Epoch 169/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0304 - accuracy: 0.9041
Epoch 170/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0198 - accuracy: 0.9178
Epoch 171/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0215 - accuracy: 0.9178
Epoch 172/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0243 - accuracy: 0.9178
Epoch 173/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0256 - accuracy: 0.9178
Epoch 174/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0239 - accuracy: 0.9178
Epoch 175/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0205 - accuracy: 0.9178
Epoch 176/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0185 - accuracy: 0.9178
Epoch 177/300
73/73 [==============================] - 0s 151us/sample - loss: 0.0261 - accuracy: 0.9178
Epoch 178/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0203 - accuracy: 0.9315
Epoch 179/300
73/73 [==============================] - 0s 152us/sample - loss: 0.0225 - accuracy: 0.9178
Epoch 180/300
73/73 [==============================] - 0s 126us/sample - loss: 0.0236 - accuracy: 0.9178
Epoch 181/300
73/73 [==============================] - 0s 137us/sample - loss: 0.0207 - accuracy: 0.9178
Epoch 182/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0218 - accuracy: 0.9178
Epoch 183/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0193 - accuracy: 0.9178
Epoch 184/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0184 - accuracy: 0.9315
Epoch 185/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0186 - accuracy: 0.9178
Epoch 186/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0233 - accuracy: 0.9178
Epoch 187/300
73/73 [==============================] - 0s 141us/sample - loss: 0.0192 - accuracy: 0.9178
Epoch 188/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0252 - accuracy: 0.9041
Epoch 189/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0246 - accuracy: 0.9178
Epoch 190/300
73/73 [==============================] - 0s 145us/sample - loss: 0.0221 - accuracy: 0.9315
Epoch 191/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0218 - accuracy: 0.9178
Epoch 192/300
73/73 [==============================] - 0s 153us/sample - loss: 0.0205 - accuracy: 0.9178
Epoch 193/300
73/73 [==============================] - 0s 142us/sample - loss: 0.0255 - accuracy: 0.9178
Epoch 194/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0202 - accuracy: 0.9178
Epoch 195/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0178 - accuracy: 0.9315
Epoch 196/300
73/73 [==============================] - 0s 145us/sample - loss: 0.0193 - accuracy: 0.9315
Epoch 197/300
73/73 [==============================] - 0s 127us/sample - loss: 0.0206 - accuracy: 0.9315
Epoch 198/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0202 - accuracy: 0.9178
Epoch 199/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0283 - accuracy: 0.9178
Epoch 200/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0263 - accuracy: 0.9178
Epoch 201/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0202 - accuracy: 0.9178
Epoch 202/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0241 - accuracy: 0.9178
Epoch 203/300
73/73 [==============================] - 0s 123us/sample - loss: 0.0231 - accuracy: 0.9315
Epoch 204/300
73/73 [==============================] - 0s 126us/sample - loss: 0.0214 - accuracy: 0.9178
Epoch 205/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0252 - accuracy: 0.9178
Epoch 206/300
73/73 [==============================] - 0s 126us/sample - loss: 0.0215 - accuracy: 0.9178
Epoch 207/300
73/73 [==============================] - 0s 127us/sample - loss: 0.0258 - accuracy: 0.9178
Epoch 208/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0239 - accuracy: 0.9178
Epoch 209/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0240 - accuracy: 0.9178
Epoch 210/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0218 - accuracy: 0.9178
Epoch 211/300
73/73 [==============================] - ETA: 0s - loss: 0.0356 - accuracy: 0.86 - 0s 138us/sample - loss: 0.0184 - accuracy: 0.9315
Epoch 212/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0298 - accuracy: 0.9178
Epoch 213/300
73/73 [==============================] - 0s 127us/sample - loss: 0.0211 - accuracy: 0.9178
Epoch 214/300
73/73 [==============================] - 0s 145us/sample - loss: 0.0238 - accuracy: 0.9178
Epoch 215/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0247 - accuracy: 0.9315
Epoch 216/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0232 - accuracy: 0.9178
Epoch 217/300
73/73 [==============================] - 0s 148us/sample - loss: 0.0230 - accuracy: 0.9178
Epoch 218/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0227 - accuracy: 0.9178
Epoch 219/300
73/73 [==============================] - 0s 137us/sample - loss: 0.0234 - accuracy: 0.9178
Epoch 220/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0213 - accuracy: 0.9178
Epoch 221/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0199 - accuracy: 0.9178
Epoch 222/300
73/73 [==============================] - 0s 124us/sample - loss: 0.0208 - accuracy: 0.9178
Epoch 223/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0222 - accuracy: 0.9178
Epoch 224/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0293 - accuracy: 0.9178
Epoch 225/300
73/73 [==============================] - 0s 123us/sample - loss: 0.0230 - accuracy: 0.9178
Epoch 226/300
73/73 [==============================] - 0s 137us/sample - loss: 0.0227 - accuracy: 0.9178
Epoch 227/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0258 - accuracy: 0.9315
Epoch 228/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0209 - accuracy: 0.9178
Epoch 229/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0219 - accuracy: 0.9178
Epoch 230/300
73/73 [==============================] - 0s 141us/sample - loss: 0.0223 - accuracy: 0.9178
Epoch 231/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0259 - accuracy: 0.9178
Epoch 232/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0231 - accuracy: 0.9178
Epoch 233/300
73/73 [==============================] - 0s 145us/sample - loss: 0.0199 - accuracy: 0.9178
Epoch 234/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0260 - accuracy: 0.9178
Epoch 235/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0195 - accuracy: 0.9178
Epoch 236/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0214 - accuracy: 0.9178
Epoch 237/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0244 - accuracy: 0.9178
Epoch 238/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0228 - accuracy: 0.9178
Epoch 239/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0214 - accuracy: 0.9178
Epoch 240/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0260 - accuracy: 0.9041
Epoch 241/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0224 - accuracy: 0.9315
Epoch 242/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0179 - accuracy: 0.9178
Epoch 243/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0210 - accuracy: 0.9178
Epoch 244/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0194 - accuracy: 0.9178
Epoch 245/300
73/73 [==============================] - ETA: 0s - loss: 9.4358e-04 - accuracy: 1.00 - 0s 134us/sample - loss: 0.0238 - accuracy: 0.9178
Epoch 246/300
73/73 [==============================] - ETA: 0s - loss: 0.0306 - accuracy: 0.93 - 0s 132us/sample - loss: 0.0246 - accuracy: 0.9178
Epoch 247/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0236 - accuracy: 0.9178
Epoch 248/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0217 - accuracy: 0.9178
Epoch 249/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0269 - accuracy: 0.9178
Epoch 250/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0212 - accuracy: 0.9178
Epoch 251/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0231 - accuracy: 0.9178
Epoch 252/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0204 - accuracy: 0.9178
Epoch 253/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0168 - accuracy: 0.9178
Epoch 254/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0236 - accuracy: 0.9178
Epoch 255/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0206 - accuracy: 0.9178
Epoch 256/300
73/73 [==============================] - 0s 126us/sample - loss: 0.0222 - accuracy: 0.9178
Epoch 257/300
73/73 [==============================] - 0s 143us/sample - loss: 0.0223 - accuracy: 0.9178
Epoch 258/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0247 - accuracy: 0.9178
Epoch 259/300
73/73 [==============================] - 0s 127us/sample - loss: 0.0229 - accuracy: 0.9178
Epoch 260/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0201 - accuracy: 0.9178
Epoch 261/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0197 - accuracy: 0.9178
Epoch 262/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0230 - accuracy: 0.9178
Epoch 263/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0188 - accuracy: 0.9178
Epoch 264/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0213 - accuracy: 0.9178
Epoch 265/300
73/73 [==============================] - 0s 124us/sample - loss: 0.0196 - accuracy: 0.9178
Epoch 266/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0225 - accuracy: 0.9178
Epoch 267/300
73/73 [==============================] - 0s 145us/sample - loss: 0.0227 - accuracy: 0.9178
Epoch 268/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0168 - accuracy: 0.9178
Epoch 269/300
73/73 [==============================] - 0s 135us/sample - loss: 0.0214 - accuracy: 0.9178
Epoch 270/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0223 - accuracy: 0.9178
Epoch 271/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0207 - accuracy: 0.9178
Epoch 272/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0225 - accuracy: 0.9178
Epoch 273/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0200 - accuracy: 0.9178
Epoch 274/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0225 - accuracy: 0.9178
Epoch 275/300
73/73 [==============================] - 0s 150us/sample - loss: 0.0271 - accuracy: 0.9178
Epoch 276/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0204 - accuracy: 0.9178
Epoch 277/300
73/73 [==============================] - 0s 138us/sample - loss: 0.0249 - accuracy: 0.9178
Epoch 278/300
73/73 [==============================] - 0s 134us/sample - loss: 0.0227 - accuracy: 0.9178
Epoch 279/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0240 - accuracy: 0.9178
Epoch 280/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0219 - accuracy: 0.9178
Epoch 281/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0257 - accuracy: 0.9041
Epoch 282/300
73/73 [==============================] - 0s 141us/sample - loss: 0.0187 - accuracy: 0.9178
Epoch 283/300
73/73 [==============================] - 0s 129us/sample - loss: 0.0199 - accuracy: 0.9178
Epoch 284/300
73/73 [==============================] - 0s 131us/sample - loss: 0.0192 - accuracy: 0.9178
Epoch 285/300
73/73 [==============================] - 0s 139us/sample - loss: 0.0205 - accuracy: 0.9178
Epoch 286/300
73/73 [==============================] - 0s 130us/sample - loss: 0.0214 - accuracy: 0.9178
Epoch 287/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0219 - accuracy: 0.9178
Epoch 288/300
73/73 [==============================] - 0s 132us/sample - loss: 0.0220 - accuracy: 0.9178
Epoch 289/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0210 - accuracy: 0.9178
Epoch 290/300
73/73 [==============================] - 0s 195us/sample - loss: 0.0199 - accuracy: 0.9178
Epoch 291/300
73/73 [==============================] - 0s 154us/sample - loss: 0.0227 - accuracy: 0.9178
Epoch 292/300
73/73 [==============================] - ETA: 0s - loss: 0.0282 - accuracy: 0.80 - 0s 150us/sample - loss: 0.0180 - accuracy: 0.9178
Epoch 293/300
73/73 [==============================] - 0s 184us/sample - loss: 0.0177 - accuracy: 0.9178
Epoch 294/300
73/73 [==============================] - 0s 144us/sample - loss: 0.0222 - accuracy: 0.9315
Epoch 295/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0180 - accuracy: 0.9178
Epoch 296/300
73/73 [==============================] - 0s 133us/sample - loss: 0.0214 - accuracy: 0.9178
Epoch 297/300
73/73 [==============================] - 0s 141us/sample - loss: 0.0206 - accuracy: 0.9178
Epoch 298/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0208 - accuracy: 0.9178
Epoch 299/300
73/73 [==============================] - 0s 128us/sample - loss: 0.0222 - accuracy: 0.9178
Epoch 300/300
73/73 [==============================] - 0s 136us/sample - loss: 0.0190 - accuracy: 0.9178
Model: "sequential_13"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense_32 (Dense) (None, 128) 4608
_________________________________________________________________
activation_32 (Activation) (None, 128) 0
_________________________________________________________________
dropout_21 (Dropout) (None, 128) 0
_________________________________________________________________
dense_33 (Dense) (None, 1) 129
_________________________________________________________________
activation_33 (Activation) (None, 1) 0
=================================================================
Total params: 4,737
Trainable params: 4,737
Non-trainable params: 0
_________________________________________________________________
Train on 74 samples
Epoch 1/300
74/74 [==============================] - 1s 7ms/sample - loss: 0.1509 - accuracy: 0.3784
Epoch 2/300
74/74 [==============================] - 0s 144us/sample - loss: 0.1408 - accuracy: 0.4189
Epoch 3/300
74/74 [==============================] - 0s 135us/sample - loss: 0.1246 - accuracy: 0.5811
Epoch 4/300
74/74 [==============================] - 0s 138us/sample - loss: 0.1236 - accuracy: 0.6351
Epoch 5/300
74/74 [==============================] - 0s 130us/sample - loss: 0.1165 - accuracy: 0.6622
Epoch 6/300
74/74 [==============================] - 0s 136us/sample - loss: 0.1111 - accuracy: 0.7027
Epoch 7/300
74/74 [==============================] - 0s 138us/sample - loss: 0.1085 - accuracy: 0.7297
Epoch 8/300
74/74 [==============================] - 0s 130us/sample - loss: 0.1057 - accuracy: 0.7973
Epoch 9/300
74/74 [==============================] - 0s 129us/sample - loss: 0.1005 - accuracy: 0.8108
Epoch 10/300
74/74 [==============================] - 0s 132us/sample - loss: 0.1018 - accuracy: 0.8243
Epoch 11/300
74/74 [==============================] - 0s 132us/sample - loss: 0.0886 - accuracy: 0.8108
Epoch 12/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0900 - accuracy: 0.8649
Epoch 13/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0827 - accuracy: 0.8784
Epoch 14/300
74/74 [==============================] - 0s 131us/sample - loss: 0.0843 - accuracy: 0.8514
Epoch 15/300
74/74 [==============================] - 0s 132us/sample - loss: 0.0771 - accuracy: 0.8784
Epoch 16/300
74/74 [==============================] - 0s 132us/sample - loss: 0.0774 - accuracy: 0.8919
Epoch 17/300
74/74 [==============================] - 0s 131us/sample - loss: 0.0703 - accuracy: 0.8784
Epoch 18/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0686 - accuracy: 0.8784
Epoch 19/300
74/74 [==============================] - 0s 144us/sample - loss: 0.0724 - accuracy: 0.8784
Epoch 20/300
74/74 [==============================] - 0s 137us/sample - loss: 0.0600 - accuracy: 0.8919
Epoch 21/300
74/74 [==============================] - 0s 136us/sample - loss: 0.0621 - accuracy: 0.8784
Epoch 22/300
74/74 [==============================] - 0s 142us/sample - loss: 0.0650 - accuracy: 0.8784
Epoch 23/300
74/74 [==============================] - 0s 137us/sample - loss: 0.0611 - accuracy: 0.8784
Epoch 24/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0568 - accuracy: 0.8784
Epoch 25/300
74/74 [==============================] - 0s 144us/sample - loss: 0.0544 - accuracy: 0.8784
Epoch 26/300
74/74 [==============================] - 0s 135us/sample - loss: 0.0543 - accuracy: 0.8919
Epoch 27/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0568 - accuracy: 0.8784
Epoch 28/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0595 - accuracy: 0.8784
Epoch 29/300
74/74 [==============================] - 0s 135us/sample - loss: 0.0570 - accuracy: 0.8784
Epoch 30/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0481 - accuracy: 0.8784
Epoch 31/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0461 - accuracy: 0.8784
Epoch 32/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0500 - accuracy: 0.8784
Epoch 33/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0429 - accuracy: 0.8919
Epoch 34/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0440 - accuracy: 0.8784
Epoch 35/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0505 - accuracy: 0.8784
Epoch 36/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0483 - accuracy: 0.8784
Epoch 37/300
74/74 [==============================] - 0s 131us/sample - loss: 0.0430 - accuracy: 0.8784
Epoch 38/300
74/74 [==============================] - 0s 132us/sample - loss: 0.0395 - accuracy: 0.8919
Epoch 39/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0417 - accuracy: 0.8784
Epoch 40/300
74/74 [==============================] - 0s 133us/sample - loss: 0.0424 - accuracy: 0.8919
Epoch 41/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0397 - accuracy: 0.8919
Epoch 42/300
74/74 [==============================] - 0s 139us/sample - loss: 0.0396 - accuracy: 0.8919
Epoch 43/300
74/74 [==============================] - 0s 133us/sample - loss: 0.0325 - accuracy: 0.8784
Epoch 44/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0402 - accuracy: 0.8919
Epoch 45/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0396 - accuracy: 0.8919
Epoch 46/300
74/74 [==============================] - 0s 133us/sample - loss: 0.0361 - accuracy: 0.8919
Epoch 47/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0377 - accuracy: 0.8919
Epoch 48/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0379 - accuracy: 0.8919
Epoch 49/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0373 - accuracy: 0.8919
Epoch 50/300
74/74 [==============================] - 0s 117us/sample - loss: 0.0379 - accuracy: 0.8919
Epoch 51/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0380 - accuracy: 0.8919
Epoch 52/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0398 - accuracy: 0.8919
Epoch 53/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0424 - accuracy: 0.8784
Epoch 54/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0390 - accuracy: 0.8919
Epoch 55/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0322 - accuracy: 0.8919
Epoch 56/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0310 - accuracy: 0.8919
Epoch 57/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0354 - accuracy: 0.8919
Epoch 58/300
74/74 [==============================] - 0s 142us/sample - loss: 0.0365 - accuracy: 0.8919
Epoch 59/300
74/74 [==============================] - 0s 132us/sample - loss: 0.0312 - accuracy: 0.8919
Epoch 60/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0396 - accuracy: 0.8919
Epoch 61/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0326 - accuracy: 0.8919
Epoch 62/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0389 - accuracy: 0.8919
Epoch 63/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0311 - accuracy: 0.8919
Epoch 64/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0341 - accuracy: 0.8919
Epoch 65/300
74/74 [==============================] - ETA: 0s - loss: 0.0291 - accuracy: 0.86 - 0s 138us/sample - loss: 0.0308 - accuracy: 0.8919
Epoch 66/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0383 - accuracy: 0.8919
Epoch 67/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0340 - accuracy: 0.8919
Epoch 68/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0321 - accuracy: 0.8919
Epoch 69/300
74/74 [==============================] - 0s 120us/sample - loss: 0.0310 - accuracy: 0.8919
Epoch 70/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0316 - accuracy: 0.8919
Epoch 71/300
74/74 [==============================] - 0s 120us/sample - loss: 0.0296 - accuracy: 0.8919
Epoch 72/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0406 - accuracy: 0.8919
Epoch 73/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0283 - accuracy: 0.8919
Epoch 74/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0326 - accuracy: 0.8919
Epoch 75/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0310 - accuracy: 0.8919
Epoch 76/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0299 - accuracy: 0.8919
Epoch 77/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0341 - accuracy: 0.8919
Epoch 78/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0316 - accuracy: 0.8919
Epoch 79/300
74/74 [==============================] - 0s 131us/sample - loss: 0.0306 - accuracy: 0.8919
Epoch 80/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0365 - accuracy: 0.8919
Epoch 81/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0367 - accuracy: 0.8919
Epoch 82/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0322 - accuracy: 0.8919
Epoch 83/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0370 - accuracy: 0.8919
Epoch 84/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0311 - accuracy: 0.8919
Epoch 85/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0341 - accuracy: 0.8919
Epoch 86/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0275 - accuracy: 0.9054
Epoch 87/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0279 - accuracy: 0.8919
Epoch 88/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0300 - accuracy: 0.8919
Epoch 89/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0319 - accuracy: 0.8919
Epoch 90/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0285 - accuracy: 0.8919
Epoch 91/300
74/74 [==============================] - 0s 119us/sample - loss: 0.0328 - accuracy: 0.8919
Epoch 92/300
74/74 [==============================] - 0s 131us/sample - loss: 0.0316 - accuracy: 0.8919
Epoch 93/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0355 - accuracy: 0.8919
Epoch 94/300
74/74 [==============================] - 0s 120us/sample - loss: 0.0332 - accuracy: 0.8919
Epoch 95/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0370 - accuracy: 0.8919
Epoch 96/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0320 - accuracy: 0.8919
Epoch 97/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0365 - accuracy: 0.8919
Epoch 98/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0280 - accuracy: 0.8919
Epoch 99/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0303 - accuracy: 0.8919
Epoch 100/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0293 - accuracy: 0.8919
Epoch 101/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0316 - accuracy: 0.8919
Epoch 102/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0334 - accuracy: 0.8919
Epoch 103/300
74/74 [==============================] - 0s 134us/sample - loss: 0.0285 - accuracy: 0.8919
Epoch 104/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0320 - accuracy: 0.8919
Epoch 105/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0290 - accuracy: 0.8919
Epoch 106/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0311 - accuracy: 0.8919
Epoch 107/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0303 - accuracy: 0.8919
Epoch 108/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0286 - accuracy: 0.8919
Epoch 109/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0297 - accuracy: 0.8919
Epoch 110/300
74/74 [==============================] - ETA: 0s - loss: 0.0160 - accuracy: 1.00 - 0s 121us/sample - loss: 0.0320 - accuracy: 0.8919
Epoch 111/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0294 - accuracy: 0.8919
Epoch 112/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0294 - accuracy: 0.8919
Epoch 113/300
74/74 [==============================] - 0s 134us/sample - loss: 0.0323 - accuracy: 0.8919
Epoch 114/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0307 - accuracy: 0.8919
Epoch 115/300
74/74 [==============================] - 0s 141us/sample - loss: 0.0320 - accuracy: 0.8919
Epoch 116/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0285 - accuracy: 0.8919
Epoch 117/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0288 - accuracy: 0.8919
Epoch 118/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0332 - accuracy: 0.8919
Epoch 119/300
74/74 [==============================] - 0s 118us/sample - loss: 0.0336 - accuracy: 0.8919
Epoch 120/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0276 - accuracy: 0.8919
Epoch 121/300
74/74 [==============================] - 0s 119us/sample - loss: 0.0306 - accuracy: 0.8919
Epoch 122/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0334 - accuracy: 0.8919
Epoch 123/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0283 - accuracy: 0.8919
Epoch 124/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0328 - accuracy: 0.8919
Epoch 125/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0317 - accuracy: 0.8919
Epoch 126/300
74/74 [==============================] - 0s 132us/sample - loss: 0.0277 - accuracy: 0.8919
Epoch 127/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0292 - accuracy: 0.8919
Epoch 128/300
74/74 [==============================] - 0s 135us/sample - loss: 0.0286 - accuracy: 0.8919
Epoch 129/300
74/74 [==============================] - 0s 132us/sample - loss: 0.0300 - accuracy: 0.8919
Epoch 130/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0374 - accuracy: 0.8919
Epoch 131/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0295 - accuracy: 0.8919
Epoch 132/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0294 - accuracy: 0.8919
Epoch 133/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0303 - accuracy: 0.8919
Epoch 134/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0311 - accuracy: 0.8919
Epoch 135/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0352 - accuracy: 0.8784
Epoch 136/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0294 - accuracy: 0.8919
Epoch 137/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0288 - accuracy: 0.8919
Epoch 138/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0333 - accuracy: 0.8919
Epoch 139/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0289 - accuracy: 0.8919
Epoch 140/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0303 - accuracy: 0.8919
Epoch 141/300
74/74 [==============================] - 0s 119us/sample - loss: 0.0319 - accuracy: 0.8919
Epoch 142/300
74/74 [==============================] - 0s 141us/sample - loss: 0.0311 - accuracy: 0.8919
Epoch 143/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0295 - accuracy: 0.8919
Epoch 144/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0299 - accuracy: 0.8919
Epoch 145/300
74/74 [==============================] - 0s 140us/sample - loss: 0.0269 - accuracy: 0.8919
Epoch 146/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0319 - accuracy: 0.8919
Epoch 147/300
74/74 [==============================] - 0s 119us/sample - loss: 0.0294 - accuracy: 0.8919
Epoch 148/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0284 - accuracy: 0.8919
Epoch 149/300
74/74 [==============================] - 0s 120us/sample - loss: 0.0281 - accuracy: 0.8919
Epoch 150/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0313 - accuracy: 0.8919
Epoch 151/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0292 - accuracy: 0.8919
Epoch 152/300
74/74 [==============================] - ETA: 0s - loss: 0.0309 - accuracy: 0.80 - 0s 115us/sample - loss: 0.0264 - accuracy: 0.8919
Epoch 153/300
74/74 [==============================] - 0s 134us/sample - loss: 0.0300 - accuracy: 0.8919
Epoch 154/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0276 - accuracy: 0.8919
Epoch 155/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0270 - accuracy: 0.8919
Epoch 156/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0329 - accuracy: 0.8919
Epoch 157/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0337 - accuracy: 0.8919
Epoch 158/300
74/74 [==============================] - 0s 118us/sample - loss: 0.0289 - accuracy: 0.8919
Epoch 159/300
74/74 [==============================] - 0s 119us/sample - loss: 0.0318 - accuracy: 0.8919
Epoch 160/300
74/74 [==============================] - 0s 120us/sample - loss: 0.0248 - accuracy: 0.8919
Epoch 161/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0293 - accuracy: 0.8919
Epoch 162/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0261 - accuracy: 0.8919
Epoch 163/300
74/74 [==============================] - 0s 119us/sample - loss: 0.0298 - accuracy: 0.8919
Epoch 164/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0330 - accuracy: 0.8919
Epoch 165/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0296 - accuracy: 0.8919
Epoch 166/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0282 - accuracy: 0.8919
Epoch 167/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0272 - accuracy: 0.8919
Epoch 168/300
74/74 [==============================] - 0s 132us/sample - loss: 0.0337 - accuracy: 0.8919
Epoch 169/300
74/74 [==============================] - 0s 118us/sample - loss: 0.0350 - accuracy: 0.8919
Epoch 170/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0339 - accuracy: 0.8919
Epoch 171/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0260 - accuracy: 0.8919
Epoch 172/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0299 - accuracy: 0.8919
Epoch 173/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0341 - accuracy: 0.8919
Epoch 174/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0308 - accuracy: 0.8919
Epoch 175/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0268 - accuracy: 0.8919
Epoch 176/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0303 - accuracy: 0.8919
Epoch 177/300
74/74 [==============================] - 0s 118us/sample - loss: 0.0279 - accuracy: 0.8919
Epoch 178/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0308 - accuracy: 0.8919
Epoch 179/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0298 - accuracy: 0.8919
Epoch 180/300
74/74 [==============================] - 0s 119us/sample - loss: 0.0282 - accuracy: 0.8919
Epoch 181/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0322 - accuracy: 0.8919
Epoch 182/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0268 - accuracy: 0.8919
Epoch 183/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0267 - accuracy: 0.8919
Epoch 184/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0300 - accuracy: 0.8919
Epoch 185/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0289 - accuracy: 0.8919
Epoch 186/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0296 - accuracy: 0.8919
Epoch 187/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0283 - accuracy: 0.8919
Epoch 188/300
74/74 [==============================] - 0s 120us/sample - loss: 0.0298 - accuracy: 0.8919
Epoch 189/300
74/74 [==============================] - ETA: 0s - loss: 0.0218 - accuracy: 0.86 - 0s 126us/sample - loss: 0.0260 - accuracy: 0.8919
Epoch 190/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0274 - accuracy: 0.8919
Epoch 191/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0280 - accuracy: 0.8919
Epoch 192/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0258 - accuracy: 0.8919
Epoch 193/300
74/74 [==============================] - ETA: 0s - loss: 0.0083 - accuracy: 0.93 - 0s 121us/sample - loss: 0.0284 - accuracy: 0.8919
Epoch 194/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0300 - accuracy: 0.8919
Epoch 195/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0269 - accuracy: 0.8919
Epoch 196/300
74/74 [==============================] - 0s 120us/sample - loss: 0.0269 - accuracy: 0.8919
Epoch 197/300
74/74 [==============================] - 0s 137us/sample - loss: 0.0308 - accuracy: 0.8919
Epoch 198/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0243 - accuracy: 0.8919
Epoch 199/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0280 - accuracy: 0.8919
Epoch 200/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0269 - accuracy: 0.9054
Epoch 201/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0312 - accuracy: 0.8919
Epoch 202/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0324 - accuracy: 0.8919
Epoch 203/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0298 - accuracy: 0.8919
Epoch 204/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0314 - accuracy: 0.8919
Epoch 205/300
74/74 [==============================] - 0s 119us/sample - loss: 0.0286 - accuracy: 0.8919
Epoch 206/300
74/74 [==============================] - 0s 132us/sample - loss: 0.0298 - accuracy: 0.8919
Epoch 207/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0320 - accuracy: 0.8919
Epoch 208/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0360 - accuracy: 0.8919
Epoch 209/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0308 - accuracy: 0.8919
Epoch 210/300
74/74 [==============================] - 0s 120us/sample - loss: 0.0280 - accuracy: 0.8919
Epoch 211/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0342 - accuracy: 0.8919
Epoch 212/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0311 - accuracy: 0.8919
Epoch 213/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0269 - accuracy: 0.8919
Epoch 214/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0285 - accuracy: 0.8919
Epoch 215/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0290 - accuracy: 0.8919
Epoch 216/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0260 - accuracy: 0.8919
Epoch 217/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0283 - accuracy: 0.8919
Epoch 218/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0307 - accuracy: 0.8919
Epoch 219/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0293 - accuracy: 0.8919
Epoch 220/300
74/74 [==============================] - 0s 135us/sample - loss: 0.0269 - accuracy: 0.8919
Epoch 221/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0306 - accuracy: 0.8919
Epoch 222/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0356 - accuracy: 0.8919
Epoch 223/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0251 - accuracy: 0.8919
Epoch 224/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0247 - accuracy: 0.8919
Epoch 225/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0276 - accuracy: 0.8919
Epoch 226/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0312 - accuracy: 0.8919
Epoch 227/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0299 - accuracy: 0.8919
Epoch 228/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0282 - accuracy: 0.8919
Epoch 229/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0289 - accuracy: 0.8919
Epoch 230/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0291 - accuracy: 0.8919
Epoch 231/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0266 - accuracy: 0.8919
Epoch 232/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0269 - accuracy: 0.8919
Epoch 233/300
74/74 [==============================] - 0s 150us/sample - loss: 0.0259 - accuracy: 0.8919
Epoch 234/300
74/74 [==============================] - 0s 136us/sample - loss: 0.0308 - accuracy: 0.8919
Epoch 235/300
74/74 [==============================] - 0s 185us/sample - loss: 0.0306 - accuracy: 0.8919
Epoch 236/300
74/74 [==============================] - 0s 142us/sample - loss: 0.0309 - accuracy: 0.8919
Epoch 237/300
74/74 [==============================] - 0s 142us/sample - loss: 0.0306 - accuracy: 0.8919
Epoch 238/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0299 - accuracy: 0.8919
Epoch 239/300
74/74 [==============================] - 0s 132us/sample - loss: 0.0265 - accuracy: 0.8919
Epoch 240/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0300 - accuracy: 0.8919
Epoch 241/300
74/74 [==============================] - 0s 135us/sample - loss: 0.0280 - accuracy: 0.8919
Epoch 242/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0277 - accuracy: 0.8919
Epoch 243/300
74/74 [==============================] - ETA: 0s - loss: 0.0272 - accuracy: 0.86 - 0s 120us/sample - loss: 0.0327 - accuracy: 0.8919
Epoch 244/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0320 - accuracy: 0.8919
Epoch 245/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0308 - accuracy: 0.8919
Epoch 246/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0283 - accuracy: 0.8919
Epoch 247/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0294 - accuracy: 0.8919
Epoch 248/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0271 - accuracy: 0.8919
Epoch 249/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0289 - accuracy: 0.8919
Epoch 250/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0293 - accuracy: 0.8919
Epoch 251/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0282 - accuracy: 0.8919
Epoch 252/300
74/74 [==============================] - 0s 133us/sample - loss: 0.0281 - accuracy: 0.8919
Epoch 253/300
74/74 [==============================] - 0s 125us/sample - loss: 0.0271 - accuracy: 0.8919
Epoch 254/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0275 - accuracy: 0.8919
Epoch 255/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0294 - accuracy: 0.8919
Epoch 256/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0303 - accuracy: 0.8919
Epoch 257/300
74/74 [==============================] - 0s 119us/sample - loss: 0.0244 - accuracy: 0.8919
Epoch 258/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0305 - accuracy: 0.8919
Epoch 259/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0296 - accuracy: 0.8919
Epoch 260/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0294 - accuracy: 0.8919
Epoch 261/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0273 - accuracy: 0.8919
Epoch 262/300
74/74 [==============================] - 0s 120us/sample - loss: 0.0314 - accuracy: 0.8919
Epoch 263/300
74/74 [==============================] - 0s 129us/sample - loss: 0.0335 - accuracy: 0.8919
Epoch 264/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0281 - accuracy: 0.8919
Epoch 265/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0293 - accuracy: 0.8919
Epoch 266/300
74/74 [==============================] - ETA: 0s - loss: 0.0364 - accuracy: 0.86 - 0s 123us/sample - loss: 0.0258 - accuracy: 0.8919
Epoch 267/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0321 - accuracy: 0.8919
Epoch 268/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0280 - accuracy: 0.8919
Epoch 269/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0319 - accuracy: 0.8919
Epoch 270/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0252 - accuracy: 0.8919
Epoch 271/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0268 - accuracy: 0.8919
Epoch 272/300
74/74 [==============================] - 0s 141us/sample - loss: 0.0303 - accuracy: 0.8919
Epoch 273/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0264 - accuracy: 0.8919
Epoch 274/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0273 - accuracy: 0.8919
Epoch 275/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0335 - accuracy: 0.8919
Epoch 276/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0312 - accuracy: 0.8919
Epoch 277/300
74/74 [==============================] - 0s 142us/sample - loss: 0.0307 - accuracy: 0.8919
Epoch 278/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0271 - accuracy: 0.8919
Epoch 279/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0254 - accuracy: 0.8919
Epoch 280/300
74/74 [==============================] - 0s 126us/sample - loss: 0.0267 - accuracy: 0.8919
Epoch 281/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0265 - accuracy: 0.8919
Epoch 282/300
74/74 [==============================] - 0s 130us/sample - loss: 0.0293 - accuracy: 0.8919
Epoch 283/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0322 - accuracy: 0.8919
Epoch 284/300
74/74 [==============================] - 0s 122us/sample - loss: 0.0256 - accuracy: 0.8919
Epoch 285/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0296 - accuracy: 0.8919
Epoch 286/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0291 - accuracy: 0.8919
Epoch 287/300
74/74 [==============================] - 0s 118us/sample - loss: 0.0297 - accuracy: 0.8919
Epoch 288/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0345 - accuracy: 0.8919
Epoch 289/300
74/74 [==============================] - 0s 123us/sample - loss: 0.0270 - accuracy: 0.8919
Epoch 290/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0287 - accuracy: 0.8919
Epoch 291/300
74/74 [==============================] - ETA: 0s - loss: 0.0515 - accuracy: 0.73 - 0s 129us/sample - loss: 0.0314 - accuracy: 0.8919
Epoch 292/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0285 - accuracy: 0.8919
Epoch 293/300
74/74 [==============================] - 0s 128us/sample - loss: 0.0303 - accuracy: 0.8919
Epoch 294/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0291 - accuracy: 0.8919
Epoch 295/300
74/74 [==============================] - 0s 120us/sample - loss: 0.0298 - accuracy: 0.8919
Epoch 296/300
74/74 [==============================] - 0s 121us/sample - loss: 0.0287 - accuracy: 0.8919
Epoch 297/300
74/74 [==============================] - 0s 140us/sample - loss: 0.0298 - accuracy: 0.8919
Epoch 298/300
74/74 [==============================] - 0s 119us/sample - loss: 0.0290 - accuracy: 0.8919
Epoch 299/300
74/74 [==============================] - 0s 127us/sample - loss: 0.0302 - accuracy: 0.8919
Epoch 300/300
74/74 [==============================] - 0s 124us/sample - loss: 0.0310 - accuracy: 0.8919
Model: "sequential_14"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense_34 (Dense) (None, 128) 4608
_________________________________________________________________
activation_34 (Activation) (None, 128) 0
_________________________________________________________________
dropout_22 (Dropout) (None, 128) 0
_________________________________________________________________
dense_35 (Dense) (None, 1) 129
_________________________________________________________________
activation_35 (Activation) (None, 1) 0
=================================================================
Total params: 4,737
Trainable params: 4,737
Non-trainable params: 0
_________________________________________________________________
Train on 75 samples
Epoch 1/300
75/75 [==============================] - 0s 4ms/sample - loss: 0.1469 - accuracy: 0.4400
Epoch 2/300
75/75 [==============================] - 0s 125us/sample - loss: 0.1418 - accuracy: 0.4800
Epoch 3/300
75/75 [==============================] - 0s 113us/sample - loss: 0.1320 - accuracy: 0.6533
Epoch 4/300
75/75 [==============================] - 0s 120us/sample - loss: 0.1309 - accuracy: 0.6533
Epoch 5/300
75/75 [==============================] - 0s 107us/sample - loss: 0.1226 - accuracy: 0.7467
Epoch 6/300
75/75 [==============================] - 0s 109us/sample - loss: 0.1153 - accuracy: 0.7733
Epoch 7/300
75/75 [==============================] - 0s 109us/sample - loss: 0.1153 - accuracy: 0.7333
Epoch 8/300
75/75 [==============================] - 0s 107us/sample - loss: 0.1117 - accuracy: 0.8267
Epoch 9/300
75/75 [==============================] - 0s 106us/sample - loss: 0.1069 - accuracy: 0.8267
Epoch 10/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0965 - accuracy: 0.8267
Epoch 11/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0983 - accuracy: 0.8000
Epoch 12/300
75/75 [==============================] - 0s 109us/sample - loss: 0.1007 - accuracy: 0.8133
Epoch 13/300
75/75 [==============================] - 0s 110us/sample - loss: 0.1000 - accuracy: 0.8133
Epoch 14/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0910 - accuracy: 0.8000
Epoch 15/300
75/75 [==============================] - 0s 105us/sample - loss: 0.0849 - accuracy: 0.8533
Epoch 16/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0848 - accuracy: 0.8533
Epoch 17/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0836 - accuracy: 0.8133
Epoch 18/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0772 - accuracy: 0.8533
Epoch 19/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0832 - accuracy: 0.8533
Epoch 20/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0777 - accuracy: 0.8533
Epoch 21/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0716 - accuracy: 0.8533
Epoch 22/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0711 - accuracy: 0.8533
Epoch 23/300
75/75 [==============================] - 0s 119us/sample - loss: 0.0730 - accuracy: 0.8533
Epoch 24/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0715 - accuracy: 0.8533
Epoch 25/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0703 - accuracy: 0.8533
Epoch 26/300
75/75 [==============================] - 0s 120us/sample - loss: 0.0724 - accuracy: 0.8533
Epoch 27/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0633 - accuracy: 0.8400
Epoch 28/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0630 - accuracy: 0.8533
Epoch 29/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0632 - accuracy: 0.8533
Epoch 30/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0665 - accuracy: 0.8533
Epoch 31/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0635 - accuracy: 0.8533
Epoch 32/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0539 - accuracy: 0.8533
Epoch 33/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0621 - accuracy: 0.8533
Epoch 34/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0615 - accuracy: 0.8533
Epoch 35/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0580 - accuracy: 0.8533
Epoch 36/300
75/75 [==============================] - 0s 104us/sample - loss: 0.0578 - accuracy: 0.8667
Epoch 37/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0513 - accuracy: 0.8533
Epoch 38/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0497 - accuracy: 0.8533
Epoch 39/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0513 - accuracy: 0.8400
Epoch 40/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0544 - accuracy: 0.8533
Epoch 41/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0531 - accuracy: 0.8533
Epoch 42/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0544 - accuracy: 0.8533
Epoch 43/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0508 - accuracy: 0.8533
Epoch 44/300
75/75 [==============================] - 0s 115us/sample - loss: 0.0519 - accuracy: 0.8533
Epoch 45/300
75/75 [==============================] - ETA: 0s - loss: 0.0583 - accuracy: 0.86 - 0s 113us/sample - loss: 0.0492 - accuracy: 0.8533
Epoch 46/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0501 - accuracy: 0.8533
Epoch 47/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0535 - accuracy: 0.8533
Epoch 48/300
75/75 [==============================] - 0s 105us/sample - loss: 0.0510 - accuracy: 0.8533
Epoch 49/300
75/75 [==============================] - 0s 115us/sample - loss: 0.0465 - accuracy: 0.8533
Epoch 50/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0525 - accuracy: 0.8533
Epoch 51/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0498 - accuracy: 0.8533
Epoch 52/300
75/75 [==============================] - 0s 125us/sample - loss: 0.0471 - accuracy: 0.8533
Epoch 53/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0505 - accuracy: 0.8533
Epoch 54/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0447 - accuracy: 0.8533
Epoch 55/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0471 - accuracy: 0.8533
Epoch 56/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0477 - accuracy: 0.8533
Epoch 57/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0448 - accuracy: 0.8533
Epoch 58/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0521 - accuracy: 0.8533
Epoch 59/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0468 - accuracy: 0.8533
Epoch 60/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0484 - accuracy: 0.8533
Epoch 61/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0429 - accuracy: 0.8533
Epoch 62/300
75/75 [==============================] - 0s 101us/sample - loss: 0.0457 - accuracy: 0.8667
Epoch 63/300
75/75 [==============================] - 0s 116us/sample - loss: 0.0435 - accuracy: 0.8533
Epoch 64/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0454 - accuracy: 0.8533
Epoch 65/300
75/75 [==============================] - ETA: 0s - loss: 0.0505 - accuracy: 0.86 - 0s 111us/sample - loss: 0.0430 - accuracy: 0.8533
Epoch 66/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0461 - accuracy: 0.8533
Epoch 67/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0461 - accuracy: 0.8667
Epoch 68/300
75/75 [==============================] - 0s 126us/sample - loss: 0.0432 - accuracy: 0.8533
Epoch 69/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0429 - accuracy: 0.8533
Epoch 70/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0392 - accuracy: 0.8533
Epoch 71/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0398 - accuracy: 0.8533
Epoch 72/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0427 - accuracy: 0.8533
Epoch 73/300
75/75 [==============================] - ETA: 0s - loss: 0.0392 - accuracy: 0.86 - 0s 112us/sample - loss: 0.0443 - accuracy: 0.8533
Epoch 74/300
75/75 [==============================] - 0s 105us/sample - loss: 0.0473 - accuracy: 0.8533
Epoch 75/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0483 - accuracy: 0.8533
Epoch 76/300
75/75 [==============================] - 0s 105us/sample - loss: 0.0430 - accuracy: 0.8533
Epoch 77/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0395 - accuracy: 0.8533
Epoch 78/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0420 - accuracy: 0.8533
Epoch 79/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0459 - accuracy: 0.8533
Epoch 80/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0353 - accuracy: 0.8533
Epoch 81/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0414 - accuracy: 0.8533
Epoch 82/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0386 - accuracy: 0.8533
Epoch 83/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0430 - accuracy: 0.8533
Epoch 84/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0380 - accuracy: 0.8533
Epoch 85/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0420 - accuracy: 0.8667
Epoch 86/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0411 - accuracy: 0.8533
Epoch 87/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0402 - accuracy: 0.8533
Epoch 88/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0429 - accuracy: 0.8533
Epoch 89/300
75/75 [==============================] - 0s 120us/sample - loss: 0.0406 - accuracy: 0.8533
Epoch 90/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0457 - accuracy: 0.8533
Epoch 91/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0404 - accuracy: 0.8667
Epoch 92/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0460 - accuracy: 0.8533
Epoch 93/300
75/75 [==============================] - 0s 105us/sample - loss: 0.0371 - accuracy: 0.8533
Epoch 94/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0398 - accuracy: 0.8533
Epoch 95/300
75/75 [==============================] - 0s 116us/sample - loss: 0.0460 - accuracy: 0.8533
Epoch 96/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0404 - accuracy: 0.8533
Epoch 97/300
75/75 [==============================] - ETA: 0s - loss: 0.0082 - accuracy: 0.93 - 0s 118us/sample - loss: 0.0384 - accuracy: 0.8533
Epoch 98/300
75/75 [==============================] - 0s 116us/sample - loss: 0.0383 - accuracy: 0.8800
Epoch 99/300
75/75 [==============================] - ETA: 0s - loss: 0.0358 - accuracy: 0.80 - 0s 113us/sample - loss: 0.0409 - accuracy: 0.8533
Epoch 100/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0447 - accuracy: 0.8533
Epoch 101/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0363 - accuracy: 0.8533
Epoch 102/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0384 - accuracy: 0.8533
Epoch 103/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0366 - accuracy: 0.8533
Epoch 104/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0469 - accuracy: 0.8533
Epoch 105/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0407 - accuracy: 0.8533
Epoch 106/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0459 - accuracy: 0.8533
Epoch 107/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0417 - accuracy: 0.8667
Epoch 108/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0416 - accuracy: 0.8533
Epoch 109/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0416 - accuracy: 0.8533
Epoch 110/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0401 - accuracy: 0.8533
Epoch 111/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0428 - accuracy: 0.8533
Epoch 112/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0399 - accuracy: 0.8533
Epoch 113/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0432 - accuracy: 0.8533
Epoch 114/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0424 - accuracy: 0.8533
Epoch 115/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0397 - accuracy: 0.8533
Epoch 116/300
75/75 [==============================] - 0s 115us/sample - loss: 0.0448 - accuracy: 0.8533
Epoch 117/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0385 - accuracy: 0.8533
Epoch 118/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0398 - accuracy: 0.8400
Epoch 119/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0365 - accuracy: 0.8533
Epoch 120/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0394 - accuracy: 0.8533
Epoch 121/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0422 - accuracy: 0.8533
Epoch 122/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0375 - accuracy: 0.8533
Epoch 123/300
75/75 [==============================] - 0s 116us/sample - loss: 0.0432 - accuracy: 0.8533
Epoch 124/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0401 - accuracy: 0.8533
Epoch 125/300
75/75 [==============================] - 0s 115us/sample - loss: 0.0428 - accuracy: 0.8533
Epoch 126/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0382 - accuracy: 0.8533
Epoch 127/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0458 - accuracy: 0.8533
Epoch 128/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0391 - accuracy: 0.8533
Epoch 129/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0393 - accuracy: 0.8533
Epoch 130/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0388 - accuracy: 0.8533
Epoch 131/300
75/75 [==============================] - 0s 105us/sample - loss: 0.0349 - accuracy: 0.8533
Epoch 132/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0389 - accuracy: 0.8533
Epoch 133/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0388 - accuracy: 0.8533
Epoch 134/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0388 - accuracy: 0.8533
Epoch 135/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0401 - accuracy: 0.8533
Epoch 136/300
75/75 [==============================] - 0s 117us/sample - loss: 0.0352 - accuracy: 0.8533
Epoch 137/300
75/75 [==============================] - 0s 116us/sample - loss: 0.0379 - accuracy: 0.8533
Epoch 138/300
75/75 [==============================] - 0s 104us/sample - loss: 0.0427 - accuracy: 0.8533
Epoch 139/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0415 - accuracy: 0.8533
Epoch 140/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0362 - accuracy: 0.8533
Epoch 141/300
75/75 [==============================] - ETA: 0s - loss: 0.0436 - accuracy: 0.93 - 0s 111us/sample - loss: 0.0397 - accuracy: 0.8533
Epoch 142/300
75/75 [==============================] - 0s 117us/sample - loss: 0.0377 - accuracy: 0.8533
Epoch 143/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0393 - accuracy: 0.8533
Epoch 144/300
75/75 [==============================] - 0s 116us/sample - loss: 0.0389 - accuracy: 0.8533
Epoch 145/300
75/75 [==============================] - 0s 116us/sample - loss: 0.0358 - accuracy: 0.8533
Epoch 146/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0394 - accuracy: 0.8533
Epoch 147/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0407 - accuracy: 0.8533
Epoch 148/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0354 - accuracy: 0.8533
Epoch 149/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0344 - accuracy: 0.8533
Epoch 150/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0364 - accuracy: 0.8533
Epoch 151/300
75/75 [==============================] - 0s 117us/sample - loss: 0.0396 - accuracy: 0.8533
Epoch 152/300
75/75 [==============================] - 0s 102us/sample - loss: 0.0358 - accuracy: 0.8533
Epoch 153/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0362 - accuracy: 0.8533
Epoch 154/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0371 - accuracy: 0.8533
Epoch 155/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0408 - accuracy: 0.8533
Epoch 156/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0373 - accuracy: 0.8533
Epoch 157/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0417 - accuracy: 0.8533
Epoch 158/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0387 - accuracy: 0.8533
Epoch 159/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0447 - accuracy: 0.8533
Epoch 160/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0387 - accuracy: 0.8533
Epoch 161/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0398 - accuracy: 0.8533
Epoch 162/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0343 - accuracy: 0.8533
Epoch 163/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0414 - accuracy: 0.8533
Epoch 164/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0365 - accuracy: 0.8533
Epoch 165/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0427 - accuracy: 0.8533
Epoch 166/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0388 - accuracy: 0.8533
Epoch 167/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0398 - accuracy: 0.8533
Epoch 168/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0389 - accuracy: 0.8533
Epoch 169/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0394 - accuracy: 0.8533
Epoch 170/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0409 - accuracy: 0.8533
Epoch 171/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0382 - accuracy: 0.8533
Epoch 172/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0399 - accuracy: 0.8533
Epoch 173/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0378 - accuracy: 0.8533
Epoch 174/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0369 - accuracy: 0.8533
Epoch 175/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0401 - accuracy: 0.8533
Epoch 176/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0370 - accuracy: 0.8667
Epoch 177/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0371 - accuracy: 0.8533
Epoch 178/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0351 - accuracy: 0.8533
Epoch 179/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0401 - accuracy: 0.8533
Epoch 180/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0387 - accuracy: 0.8533
Epoch 181/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0366 - accuracy: 0.8533
Epoch 182/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0397 - accuracy: 0.8533
Epoch 183/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0377 - accuracy: 0.8533
Epoch 184/300
75/75 [==============================] - 0s 117us/sample - loss: 0.0365 - accuracy: 0.8533
Epoch 185/300
75/75 [==============================] - 0s 116us/sample - loss: 0.0412 - accuracy: 0.8533
Epoch 186/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0408 - accuracy: 0.8533
Epoch 187/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0374 - accuracy: 0.8533
Epoch 188/300
75/75 [==============================] - 0s 104us/sample - loss: 0.0390 - accuracy: 0.8667
Epoch 189/300
75/75 [==============================] - ETA: 0s - loss: 0.0380 - accuracy: 0.93 - 0s 112us/sample - loss: 0.0394 - accuracy: 0.8533
Epoch 190/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0379 - accuracy: 0.8533
Epoch 191/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0383 - accuracy: 0.8533
Epoch 192/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0376 - accuracy: 0.8533
Epoch 193/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0388 - accuracy: 0.8533
Epoch 194/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0373 - accuracy: 0.8533
Epoch 195/300
75/75 [==============================] - 0s 104us/sample - loss: 0.0373 - accuracy: 0.8533
Epoch 196/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0354 - accuracy: 0.8533
Epoch 197/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0357 - accuracy: 0.8533
Epoch 198/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0434 - accuracy: 0.8533
Epoch 199/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0369 - accuracy: 0.8533
Epoch 200/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0339 - accuracy: 0.8533
Epoch 201/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0363 - accuracy: 0.8533
Epoch 202/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0383 - accuracy: 0.8533
Epoch 203/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0346 - accuracy: 0.8533
Epoch 204/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0360 - accuracy: 0.8533
Epoch 205/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0367 - accuracy: 0.8533
Epoch 206/300
75/75 [==============================] - 0s 115us/sample - loss: 0.0396 - accuracy: 0.8533
Epoch 207/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0358 - accuracy: 0.8533
Epoch 208/300
75/75 [==============================] - 0s 121us/sample - loss: 0.0409 - accuracy: 0.8533
Epoch 209/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0427 - accuracy: 0.8533
Epoch 210/300
75/75 [==============================] - 0s 115us/sample - loss: 0.0396 - accuracy: 0.8533
Epoch 211/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0381 - accuracy: 0.8667
Epoch 212/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0415 - accuracy: 0.8533
Epoch 213/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0371 - accuracy: 0.8533
Epoch 214/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0383 - accuracy: 0.8533
Epoch 215/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0419 - accuracy: 0.8533
Epoch 216/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0370 - accuracy: 0.8533
Epoch 217/300
75/75 [==============================] - 0s 119us/sample - loss: 0.0328 - accuracy: 0.8533
Epoch 218/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0357 - accuracy: 0.8533
Epoch 219/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0347 - accuracy: 0.8533
Epoch 220/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0365 - accuracy: 0.8533
Epoch 221/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0388 - accuracy: 0.8533
Epoch 222/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0378 - accuracy: 0.8533
Epoch 223/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0388 - accuracy: 0.8533
Epoch 224/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0353 - accuracy: 0.8533
Epoch 225/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0394 - accuracy: 0.8533
Epoch 226/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0377 - accuracy: 0.8533
Epoch 227/300
75/75 [==============================] - ETA: 0s - loss: 0.0099 - accuracy: 0.93 - 0s 114us/sample - loss: 0.0383 - accuracy: 0.8533
Epoch 228/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0407 - accuracy: 0.8533
Epoch 229/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0414 - accuracy: 0.8533
Epoch 230/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0363 - accuracy: 0.8533
Epoch 231/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0399 - accuracy: 0.8533
Epoch 232/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0403 - accuracy: 0.8533
Epoch 233/300
75/75 [==============================] - 0s 104us/sample - loss: 0.0390 - accuracy: 0.8533
Epoch 234/300
75/75 [==============================] - 0s 118us/sample - loss: 0.0387 - accuracy: 0.8533
Epoch 235/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0421 - accuracy: 0.8533
Epoch 236/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0370 - accuracy: 0.8533
Epoch 237/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0400 - accuracy: 0.8533
Epoch 238/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0360 - accuracy: 0.8533
Epoch 239/300
75/75 [==============================] - 0s 115us/sample - loss: 0.0406 - accuracy: 0.8533
Epoch 240/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0393 - accuracy: 0.8533
Epoch 241/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0392 - accuracy: 0.8533
Epoch 242/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0361 - accuracy: 0.8533
Epoch 243/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0372 - accuracy: 0.8533
Epoch 244/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0374 - accuracy: 0.8533
Epoch 245/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0411 - accuracy: 0.8533
Epoch 246/300
75/75 [==============================] - 0s 118us/sample - loss: 0.0361 - accuracy: 0.8533
Epoch 247/300
75/75 [==============================] - 0s 118us/sample - loss: 0.0358 - accuracy: 0.8533
Epoch 248/300
75/75 [==============================] - 0s 122us/sample - loss: 0.0376 - accuracy: 0.8533
Epoch 249/300
75/75 [==============================] - 0s 118us/sample - loss: 0.0350 - accuracy: 0.8533
Epoch 250/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0382 - accuracy: 0.8533
Epoch 251/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0345 - accuracy: 0.8533
Epoch 252/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0369 - accuracy: 0.8533
Epoch 253/300
75/75 [==============================] - 0s 114us/sample - loss: 0.0374 - accuracy: 0.8533
Epoch 254/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0388 - accuracy: 0.8533
Epoch 255/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0390 - accuracy: 0.8533
Epoch 256/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0410 - accuracy: 0.8533
Epoch 257/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0377 - accuracy: 0.8533
Epoch 258/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0366 - accuracy: 0.8533
Epoch 259/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0376 - accuracy: 0.8533
Epoch 260/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0406 - accuracy: 0.8533
Epoch 261/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0366 - accuracy: 0.8533
Epoch 262/300
75/75 [==============================] - 0s 104us/sample - loss: 0.0351 - accuracy: 0.8533
Epoch 263/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0352 - accuracy: 0.8533
Epoch 264/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0343 - accuracy: 0.8533
Epoch 265/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0361 - accuracy: 0.8533
Epoch 266/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0365 - accuracy: 0.8533
Epoch 267/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0387 - accuracy: 0.8533
Epoch 268/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0388 - accuracy: 0.8533
Epoch 269/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0366 - accuracy: 0.8533
Epoch 270/300
75/75 [==============================] - 0s 119us/sample - loss: 0.0351 - accuracy: 0.8533
Epoch 271/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0392 - accuracy: 0.8533
Epoch 272/300
75/75 [==============================] - 0s 121us/sample - loss: 0.0377 - accuracy: 0.8533
Epoch 273/300
75/75 [==============================] - 0s 116us/sample - loss: 0.0390 - accuracy: 0.8533
Epoch 274/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0367 - accuracy: 0.8533
Epoch 275/300
75/75 [==============================] - 0s 115us/sample - loss: 0.0407 - accuracy: 0.8533
Epoch 276/300
75/75 [==============================] - 0s 105us/sample - loss: 0.0405 - accuracy: 0.8533
Epoch 277/300
75/75 [==============================] - ETA: 0s - loss: 0.0699 - accuracy: 0.80 - 0s 115us/sample - loss: 0.0391 - accuracy: 0.8533
Epoch 278/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0367 - accuracy: 0.8533
Epoch 279/300
75/75 [==============================] - 0s 113us/sample - loss: 0.0373 - accuracy: 0.8533
Epoch 280/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0370 - accuracy: 0.8533
Epoch 281/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0417 - accuracy: 0.8533
Epoch 282/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0404 - accuracy: 0.8533
Epoch 283/300
75/75 [==============================] - 0s 107us/sample - loss: 0.0365 - accuracy: 0.8533
Epoch 284/300
75/75 [==============================] - 0s 115us/sample - loss: 0.0386 - accuracy: 0.8533
Epoch 285/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0365 - accuracy: 0.8533
Epoch 286/300
75/75 [==============================] - 0s 117us/sample - loss: 0.0353 - accuracy: 0.8533
Epoch 287/300
75/75 [==============================] - 0s 111us/sample - loss: 0.0384 - accuracy: 0.8533
Epoch 288/300
75/75 [==============================] - 0s 109us/sample - loss: 0.0373 - accuracy: 0.8533
Epoch 289/300
75/75 [==============================] - 0s 117us/sample - loss: 0.0388 - accuracy: 0.8533
Epoch 290/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0398 - accuracy: 0.8533
Epoch 291/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0396 - accuracy: 0.8533
Epoch 292/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0375 - accuracy: 0.8533
Epoch 293/300
75/75 [==============================] - 0s 112us/sample - loss: 0.0355 - accuracy: 0.8533
Epoch 294/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0371 - accuracy: 0.8533
Epoch 295/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0339 - accuracy: 0.8533
Epoch 296/300
75/75 [==============================] - 0s 120us/sample - loss: 0.0371 - accuracy: 0.8533
Epoch 297/300
75/75 [==============================] - 0s 108us/sample - loss: 0.0354 - accuracy: 0.8533
Epoch 298/300
75/75 [==============================] - 0s 110us/sample - loss: 0.0364 - accuracy: 0.8533
Epoch 299/300
75/75 [==============================] - 0s 106us/sample - loss: 0.0370 - accuracy: 0.8533
Epoch 300/300
75/75 [==============================] - 0s 103us/sample - loss: 0.0391 - accuracy: 0.8533
print ('Average f1 score', np.mean(test_F1))
print ('Average Run time', np.mean(time_k))Average f1 score 0.5851851851851851
Average Run time 3.6827285289764404
We built an LSTM Classifier on the sequences to compare the accuracy.
X = darpa_data['seq']
encoded_X = np.ndarray(shape=(len(X),), dtype=list)
for i in range(0,len(X)):
encoded_X[i]=X.iloc[i].split("~")max_seq_length = np.max(darpa_data['seqlen'])
encoded_X = tf.keras.preprocessing.sequence.pad_sequences(encoded_X, maxlen=max_seq_length)kfold = 3
random_state = 11
test_F1 = np.zeros(kfold)
time_k = np.zeros(kfold)
epochs = 50
batch_size = 15
skf = StratifiedKFold(n_splits=kfold, shuffle=True, random_state=random_state)
k = 0
for train_index, test_index in skf.split(encoded_X, y):
X_train, X_test = encoded_X[train_index], encoded_X[test_index]
y_train, y_test = y[train_index], y[test_index]
embedding_vecor_length = 32
top_words=50
model = Sequential()
model.add(Embedding(top_words, embedding_vecor_length, input_length=max_seq_length))
model.add(LSTM(32))
model.add(Dense(1))
model.add(Activation('sigmoid'))
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
model.summary()
start_time = time.time()
model.fit(X_train, y_train, epochs=epochs, batch_size=batch_size, verbose=1)
end_time=time.time()
time_k[k]=end_time-start_time
y_pred = model.predict_proba(X_test).round().astype(int)
y_train_pred=model.predict_proba(X_train).round().astype(int)
test_F1[k]=sklearn.metrics.f1_score(y_test, y_pred)
k+=1Model: "sequential_24"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
embedding_9 (Embedding) (None, 1773, 32) 1600
_________________________________________________________________
lstm_9 (LSTM) (None, 32) 8320
_________________________________________________________________
dense_44 (Dense) (None, 1) 33
_________________________________________________________________
activation_44 (Activation) (None, 1) 0
=================================================================
Total params: 9,953
Trainable params: 9,953
Non-trainable params: 0
_________________________________________________________________
Train on 73 samples
Epoch 1/50
73/73 [==============================] - 5s 71ms/sample - loss: 0.6829 - accuracy: 0.8493
Epoch 2/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.6532 - accuracy: 0.8904
Epoch 3/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.6164 - accuracy: 0.8904
Epoch 4/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.5658 - accuracy: 0.8904
Epoch 5/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.4744 - accuracy: 0.8904
Epoch 6/50
73/73 [==============================] - 3s 46ms/sample - loss: 0.3893 - accuracy: 0.8904
Epoch 7/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3459 - accuracy: 0.8904
Epoch 8/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3521 - accuracy: 0.8904
Epoch 9/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3522 - accuracy: 0.8904
Epoch 10/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3512 - accuracy: 0.8904
Epoch 11/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3452 - accuracy: 0.8904
Epoch 12/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3444 - accuracy: 0.8904
Epoch 13/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3436 - accuracy: 0.8904
Epoch 14/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3431 - accuracy: 0.8904
Epoch 15/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3425 - accuracy: 0.8904
Epoch 16/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3434 - accuracy: 0.8904
Epoch 17/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3402 - accuracy: 0.8904
Epoch 18/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3400 - accuracy: 0.8904
Epoch 19/50
73/73 [==============================] - 3s 45ms/sample - loss: 0.3378 - accuracy: 0.8904
Epoch 20/50
73/73 [==============================] - 3s 46ms/sample - loss: 0.3365 - accuracy: 0.8904
Epoch 21/50
73/73 [==============================] - 3s 45ms/sample - loss: 0.3347 - accuracy: 0.8904
Epoch 22/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3304 - accuracy: 0.8904
Epoch 23/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3252 - accuracy: 0.8904
Epoch 24/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3217 - accuracy: 0.8904
Epoch 25/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.3105 - accuracy: 0.8904
Epoch 26/50
73/73 [==============================] - 3s 43ms/sample - loss: 0.2963 - accuracy: 0.8904
Epoch 27/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.2876 - accuracy: 0.8904
Epoch 28/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.2561 - accuracy: 0.8904
Epoch 29/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.2379 - accuracy: 0.8904
Epoch 30/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.2244 - accuracy: 0.8904
Epoch 31/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.2243 - accuracy: 0.9041
Epoch 32/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.2195 - accuracy: 0.9178
Epoch 33/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1969 - accuracy: 0.9315
Epoch 34/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.2076 - accuracy: 0.8767
Epoch 35/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.2151 - accuracy: 0.8767
Epoch 36/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1920 - accuracy: 0.9041
Epoch 37/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1963 - accuracy: 0.9041
Epoch 38/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.2015 - accuracy: 0.9178
Epoch 39/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1899 - accuracy: 0.8767
Epoch 40/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1780 - accuracy: 0.9178
Epoch 41/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1784 - accuracy: 0.9315
Epoch 42/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1646 - accuracy: 0.9315
Epoch 43/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1633 - accuracy: 0.9315
Epoch 44/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1607 - accuracy: 0.9315
Epoch 45/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1527 - accuracy: 0.9315
Epoch 46/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1644 - accuracy: 0.9315
Epoch 47/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1660 - accuracy: 0.9178
Epoch 48/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1487 - accuracy: 0.9178
Epoch 49/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1992 - accuracy: 0.9315
Epoch 50/50
73/73 [==============================] - 3s 44ms/sample - loss: 0.1352 - accuracy: 0.9589
Model: "sequential_25"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
embedding_10 (Embedding) (None, 1773, 32) 1600
_________________________________________________________________
lstm_10 (LSTM) (None, 32) 8320
_________________________________________________________________
dense_45 (Dense) (None, 1) 33
_________________________________________________________________
activation_45 (Activation) (None, 1) 0
=================================================================
Total params: 9,953
Trainable params: 9,953
Non-trainable params: 0
_________________________________________________________________
Train on 74 samples
Epoch 1/50
74/74 [==============================] - 5s 71ms/sample - loss: 0.6728 - accuracy: 0.8649
Epoch 2/50
74/74 [==============================] - 3s 43ms/sample - loss: 0.6344 - accuracy: 0.8649
Epoch 3/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.5765 - accuracy: 0.8784
Epoch 4/50
74/74 [==============================] - 3s 43ms/sample - loss: 0.4936 - accuracy: 0.8784
Epoch 5/50
74/74 [==============================] - 3s 45ms/sample - loss: 0.3903 - accuracy: 0.8784
Epoch 6/50
74/74 [==============================] - 3s 45ms/sample - loss: 0.3818 - accuracy: 0.8784
Epoch 7/50
74/74 [==============================] - 3s 45ms/sample - loss: 0.3885 - accuracy: 0.8784
Epoch 8/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3802 - accuracy: 0.8784
Epoch 9/50
74/74 [==============================] - 3s 43ms/sample - loss: 0.3717 - accuracy: 0.8784
Epoch 10/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3697 - accuracy: 0.8784
Epoch 11/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3696 - accuracy: 0.8784
Epoch 12/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3687 - accuracy: 0.8784
Epoch 13/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3681 - accuracy: 0.8784
Epoch 14/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3667 - accuracy: 0.8784
Epoch 15/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3658 - accuracy: 0.8784
Epoch 16/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3656 - accuracy: 0.8784
Epoch 17/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3639 - accuracy: 0.8784
Epoch 18/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3611 - accuracy: 0.8784
Epoch 19/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3577 - accuracy: 0.8784
Epoch 20/50
74/74 [==============================] - 3s 43ms/sample - loss: 0.3554 - accuracy: 0.8784
Epoch 21/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3533 - accuracy: 0.8784
Epoch 22/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3476 - accuracy: 0.8784
Epoch 23/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3439 - accuracy: 0.8784
Epoch 24/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3297 - accuracy: 0.8784
Epoch 25/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.3186 - accuracy: 0.8784
Epoch 26/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2966 - accuracy: 0.8784
Epoch 27/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2752 - accuracy: 0.8784
Epoch 28/50
74/74 [==============================] - 3s 43ms/sample - loss: 0.2624 - accuracy: 0.8784
Epoch 29/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2652 - accuracy: 0.8919
Epoch 30/50
74/74 [==============================] - 3s 43ms/sample - loss: 0.2547 - accuracy: 0.9054
Epoch 31/50
74/74 [==============================] - 3s 43ms/sample - loss: 0.2679 - accuracy: 0.9054
Epoch 32/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2486 - accuracy: 0.8919
Epoch 33/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2146 - accuracy: 0.9054
Epoch 34/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2486 - accuracy: 0.9189
Epoch 35/50
74/74 [==============================] - 3s 43ms/sample - loss: 0.2169 - accuracy: 0.9459
Epoch 36/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2312 - accuracy: 0.8919
Epoch 37/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.1977 - accuracy: 0.9459
Epoch 38/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2101 - accuracy: 0.9459
Epoch 39/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2023 - accuracy: 0.9189
Epoch 40/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2046 - accuracy: 0.9324
Epoch 41/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.1890 - accuracy: 0.9459
Epoch 42/50
74/74 [==============================] - 3s 45ms/sample - loss: 0.1811 - accuracy: 0.9459
Epoch 43/50
74/74 [==============================] - 3s 45ms/sample - loss: 0.1917 - accuracy: 0.9459
Epoch 44/50
74/74 [==============================] - 3s 45ms/sample - loss: 0.1872 - accuracy: 0.9459
Epoch 45/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.1658 - accuracy: 0.9459
Epoch 46/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.1739 - accuracy: 0.9459
Epoch 47/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.1645 - accuracy: 0.9459
Epoch 48/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.1878 - accuracy: 0.9459
Epoch 49/50
74/74 [==============================] - 3s 43ms/sample - loss: 0.1841 - accuracy: 0.9595
Epoch 50/50
74/74 [==============================] - 3s 44ms/sample - loss: 0.2039 - accuracy: 0.8919
Model: "sequential_26"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
embedding_11 (Embedding) (None, 1773, 32) 1600
_________________________________________________________________
lstm_11 (LSTM) (None, 32) 8320
_________________________________________________________________
dense_46 (Dense) (None, 1) 33
_________________________________________________________________
activation_46 (Activation) (None, 1) 0
=================================================================
Total params: 9,953
Trainable params: 9,953
Non-trainable params: 0
_________________________________________________________________
Train on 75 samples
Epoch 1/50
75/75 [==============================] - 5s 66ms/sample - loss: 0.6830 - accuracy: 0.7333
Epoch 2/50
75/75 [==============================] - 3s 42ms/sample - loss: 0.6459 - accuracy: 0.8800
Epoch 3/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.6046 - accuracy: 0.8800
Epoch 4/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.5368 - accuracy: 0.8800
Epoch 5/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.4176 - accuracy: 0.8800
Epoch 6/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3610 - accuracy: 0.8800
Epoch 7/50
75/75 [==============================] - 3s 42ms/sample - loss: 0.3993 - accuracy: 0.8800
Epoch 8/50
75/75 [==============================] - 3s 42ms/sample - loss: 0.3872 - accuracy: 0.8800
Epoch 9/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3789 - accuracy: 0.8800
Epoch 10/50
75/75 [==============================] - 3s 42ms/sample - loss: 0.3725 - accuracy: 0.8800
Epoch 11/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3679 - accuracy: 0.8800
Epoch 12/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3678 - accuracy: 0.8800
Epoch 13/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3681 - accuracy: 0.8800
Epoch 14/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3682 - accuracy: 0.8800
Epoch 15/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3669 - accuracy: 0.8800
Epoch 16/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3677 - accuracy: 0.8800
Epoch 17/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3657 - accuracy: 0.8800
Epoch 18/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3654 - accuracy: 0.8800
Epoch 19/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3669 - accuracy: 0.8800
Epoch 20/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3650 - accuracy: 0.8800
Epoch 21/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3649 - accuracy: 0.8800
Epoch 22/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3650 - accuracy: 0.8800
Epoch 23/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3633 - accuracy: 0.8800
Epoch 24/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3620 - accuracy: 0.8800
Epoch 25/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3611 - accuracy: 0.8800
Epoch 26/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3595 - accuracy: 0.8800
Epoch 27/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3627 - accuracy: 0.8800
Epoch 28/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3559 - accuracy: 0.8800
Epoch 29/50
75/75 [==============================] - 3s 42ms/sample - loss: 0.3512 - accuracy: 0.8800
Epoch 30/50
75/75 [==============================] - 3s 42ms/sample - loss: 0.3507 - accuracy: 0.8800
Epoch 31/50
75/75 [==============================] - 3s 43ms/sample - loss: 0.3392 - accuracy: 0.8800
Epoch 32/50
75/75 [==============================] - 3s 42ms/sample - loss: 0.3340 - accuracy: 0.8800
Epoch 33/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3115 - accuracy: 0.8800
Epoch 34/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2962 - accuracy: 0.8800
Epoch 35/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2863 - accuracy: 0.8800
Epoch 36/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2715 - accuracy: 0.8800
Epoch 37/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2471 - accuracy: 0.8800
Epoch 38/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.3184 - accuracy: 0.8800
Epoch 39/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2991 - accuracy: 0.8800
Epoch 40/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2846 - accuracy: 0.8800
Epoch 41/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2525 - accuracy: 0.8800
Epoch 42/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2494 - accuracy: 0.8800
Epoch 43/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2456 - accuracy: 0.8800
Epoch 44/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2356 - accuracy: 0.8800
Epoch 45/50
75/75 [==============================] - 3s 42ms/sample - loss: 0.2281 - accuracy: 0.9067
Epoch 46/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2207 - accuracy: 0.9067
Epoch 47/50
75/75 [==============================] - 3s 42ms/sample - loss: 0.2165 - accuracy: 0.8800
Epoch 48/50
75/75 [==============================] - 3s 42ms/sample - loss: 0.2136 - accuracy: 0.8933
Epoch 49/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2141 - accuracy: 0.9067
Epoch 50/50
75/75 [==============================] - 3s 41ms/sample - loss: 0.2053 - accuracy: 0.9067
print ('Average f1 score', np.mean(test_F1))
print ('Average Run time', np.mean(time_k))Average f1 score 0.5309941520467837
Average Run time 161.5511829853058
We find that the LSTM classifier gives an F1 score of 0. This may be improved by changing the model. However, we find that the SGT embedding could work with a small and unbalanced data without the need of a complicated classifier model.
LSTM models typically require more data for training and also has significantly more computation time. The LSTM model above took 425.6 secs while the MLP model took just 9.1 secs.
Apply the algorithm on a sequence BBACACAABA. The parts of SGT, W(0) and W(\kappa), in Algorithm 1 & 2 in [1], and the resulting SGT estimate will be (line-by-line execution of main.R):
alphabet_set <- c("A", "B", "C") # Denoted by variable V in [1]
seq <- "BBACACAABA"
kappa <- 5
###### Algorithm 1 ######
sgt_parts_alg1 <- f_sgt_parts(sequence = seq, kappa = kappa, alphabet_set_size = length(alphabet_set))
print(sgt_parts_alg1)
Result
$W0
A B C
A 10 4 3
B 11 3 4
C 7 2 1
$W_kappa
A B C
A 0.006874761 6.783349e-03 1.347620e-02
B 0.013521602 6.737947e-03 4.570791e-05
C 0.013521604 3.059162e-07 4.539993e-05
sgt <- f_SGT(W_kappa = sgt_parts_alg1$W_kappa, W0 = sgt_parts_alg1$W0,
Len = sgt_parts_alg1$Len, kappa = kappa) # Set Len = NULL for length-sensitive SGT.
print(sgt)
Result
A B C
A 0.3693614 0.44246287 0.5376371
B 0.4148844 0.46803816 0.1627745
C 0.4541361 0.06869332 0.2144920
Similarly, the execution for Algorithm-2 is shown in main.R.
Open file main.R and execute line-by-line to understand the process. In this sample execution, we present SGT estimation from either of the two algorithms presented in [1]. The first part is for understanding the SGT computation process.
In the next part we demonstrate sequence clustering using SGT on a synthesized sample dataset. The sequence lengths in the dataset ranges between (45, 711) with a uniform distribution (hence, average length is ~365). Similar sequences in the dataset has some similar patterns, in turn common substrings. These common substrings can be of any length. Also, the order of the instances of these substrings is arbitrary and random in different sequences. For example, the following two sequences have common patterns. One common subtring in both is QZTA which is present arbitrarily in both sequences. The two sequences have other common substrings as well. Other than these commonlities there are significant amount of noise present in the sequences. On average, about 40% of the letters in all sequences in the dataset are noise.
AKQZTAEEYTDZUXXIRZSTAYFUIXCPDZUXMCSMEMVDVGMTDRDDEJWNDGDPSVPKJHKQBRKMXHHNLUBXBMHISQ
WEHGXGDDCADPVKESYQXGRLRZSTAYFUOQZTAWTBRKMXHHNWYRYBRKMXHHNPRNRBRKMXHHNPBMHIUSVXBMHI
WXQRZSTAYFUCWRZSTAYFUJEJDZUXPUEMVDVGMTOHUDZUXLOQSKESYQXGRCTLBRKMXHHNNJZDZUXTFWZKES
YQXGRUATSNDGDPWEBNIQZMBNIQKESYQXGRSZTTPTZWRMEMVDVGMTAPBNIRPSADZUXJTEDESOKPTLJEMZTD
LUIPSMZTDLUIWYDELISBRKMXHHNMADEDXKESYQXGRWEFRZSTAYFUDNDGDPKYEKPTSXMKNDGDPUTIQJHKSD
ZUXVMZTDLUINFNDGDPMQZTAPPKBMHIUQIUBMHIEKKJHK
SDBRKMXHHNRATBMHIYDZUXMTRMZTDLUIEKDEIBQZTAZOAMZTDLUILHGXGDDCAZEXJHKTDOOHGXGDDCAKZH
NEMVDVGMTIHZXDEROEQDEGZPPTDBCLBMHIJMMKESYQXGRGDPTNBRKMXHHNGCBYNDGDPKMWKBMHIDQDZUXI
HKVBMHINQZTAHBRKMXHHNIRBRKMXHHNDISDZUXWBOYEMVDVGMTNTAQZTA
Identifying similar sequences with good accuracy, and also low false positives (calling sequences similar when they are not) is difficult in such situations due to,
-
Different lengths of the sequences: due to different lengths figuring out that two sequences have same inherent pattern is not straightforward. Normalizing the pattern features by the sequence length is a non-trivial problem.
-
Commonalities are not in order: as shown in the above example sequences, the common substrings are anywhere. This makes methods such as alignment-based approaches infeasible.
-
Significant amount of noise: a good amount of noise is a nemesis to most sequence similarity algorithms. It often results into high false positives.
The dataset here is a good example for the above challenges. We run clustering on the dataset in main.R. The sequences in the dataset are from 5 (=K) clusters. We use this ground truth about the number of clusters as input to our execution below. Although, in reality, the true number of clusters is unknown for a data, here we are demonstrating the SGT implementation. Regardless, using the random search procedure discussed in Sec.SGT-ALGORITHM in [1], we could find the number of clusters as equal to 5. For simplicity it has been kept out of this demonstration.
Other state-of-the-art sequence clustering methods had significantly poorer performance even with the number of true clusters (K=5). HMM had good performance but significantly higher computation time.
## The dataset contains all roman letters, A-Z.
dataset <- read.csv("dataset.csv", header = T, stringsAsFactors = F)
sgt_parts_sequences_in_dataset <- f_SGT_for_each_sequence_in_dataset(sequence_dataset = dataset,
kappa = 5, alphabet_set = LETTERS,
spp = NULL, sgt_using_alphabet_positions = T)
input_data <- f_create_input_kmeans(all_seq_sgt_parts = sgt_parts_sequences_in_dataset,
length_normalize = T,
alphabet_set_size = 26,
kappa = 5, trace = TRUE,
inv.powered = T)
K = 5
clustering_output <- f_kmeans(input_data = input_data, K = K, alphabet_set_size = 26, trace = T)
cc <- f_clustering_accuracy(actual = c(strtoi(dataset[,1])), pred = c(clustering_output$class), K = K, type = "f1")
print(cc)
Result
$cc
Confusion Matrix and Statistics
Reference
Prediction a b c d e
a 50 0 0 0 0
b 0 66 0 0 0
c 0 0 60 0 0
d 0 0 0 55 0
e 0 0 0 0 68
Overall Statistics
Accuracy : 1
95% CI : (0.9877, 1)
No Information Rate : 0.2274
P-Value [Acc > NIR] : < 2.2e-16
Kappa : 1
Mcnemar's Test P-Value : NA
Statistics by Class:
Class: a Class: b Class: c Class: d Class: e
Sensitivity 1.0000 1.0000 1.0000 1.0000 1.0000
Specificity 1.0000 1.0000 1.0000 1.0000 1.0000
Pos Pred Value 1.0000 1.0000 1.0000 1.0000 1.0000
Neg Pred Value 1.0000 1.0000 1.0000 1.0000 1.0000
Prevalence 0.1672 0.2207 0.2007 0.1839 0.2274
Detection Rate 0.1672 0.2207 0.2007 0.1839 0.2274
Detection Prevalence 0.1672 0.2207 0.2007 0.1839 0.2274
Balanced Accuracy 1.0000 1.0000 1.0000 1.0000 1.0000
$F1
F1
1
As we can see the clustering result is accurate with no false-positives. The f1-score is 1.0.
Note: Do not run function
f_clustering_accuracywhenKis larger (> 7), because it does a permutation operation which will become expensive.
For demonstrating PCA on SGT for dimension reduction and then performing clustering, we added another code snippet. PCA becomes more important on datasets where SGT's are sparse. A sparse SGT is present when the alphabet set is large but the observed sequences contain only a few of those alphabets. For example, the alphabet set for sequence dataset of music listening history will have thousands to millions of songs, but a single sequence will have only a few of them
######## Clustering on Principal Components of SGT features ########
num_pcs <- 5 # Number of principal components we want
input_data_pcs <- f_pcs(input_data = input_data, PCs = num_pcs)$input_data_pcs
clustering_output_pcs <- f_kmeans(input_data = input_data_pcs, K = K, alphabet_set_size = sqrt(num_pcs), trace = F)
cc <- f_clustering_accuracy(actual = c(strtoi(dataset[,1])), pred = c(clustering_output_pcs$class), K = K, type = "f1")
print(cc)
Result
$cc
Confusion Matrix and Statistics
Reference
Prediction a b c d e
a 50 0 0 0 0
b 0 66 0 0 0
c 0 0 60 0 0
d 0 0 0 55 0
e 0 0 0 0 68
Overall Statistics
Accuracy : 1
95% CI : (0.9877, 1)
No Information Rate : 0.2274
P-Value [Acc > NIR] : < 2.2e-16
Kappa : 1
Mcnemar's Test P-Value : NA
Statistics by Class:
Class: a Class: b Class: c Class: d Class: e
Sensitivity 1.0000 1.0000 1.0000 1.0000 1.0000
Specificity 1.0000 1.0000 1.0000 1.0000 1.0000
Pos Pred Value 1.0000 1.0000 1.0000 1.0000 1.0000
Neg Pred Value 1.0000 1.0000 1.0000 1.0000 1.0000
Prevalence 0.1672 0.2207 0.2007 0.1839 0.2274
Detection Rate 0.1672 0.2207 0.2007 0.1839 0.2274
Detection Prevalence 0.1672 0.2207 0.2007 0.1839 0.2274
Balanced Accuracy 1.0000 1.0000 1.0000 1.0000 1.0000
$F1
F1
1
The clustering result remains accurate upon clustering the PCs on the SGT of sequences.
- Simplicity: SGT's is simple to implement. There is no numerical optimization or other solution search algorithm required to estimate SGT. This makes it deterministic and powerful.
- Length sensitive: The length sensitive version of SGT can be easily tried by changing the marked arguments in
main.R.
- Small alphabet set: If the alphabet set is small (< 4), SGT's performance may not be good. This is because the feature space becomes too small.
- Faster implementation: The provided code is a research level code, not optimized for the best of speed. Significant speed improvements can be made, e.g. multithreading the SGT estimation for sequences in a dataset.
Python implementation: Please refer to
Thanks to Yassine for providing the Python implementation.