-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy paththesis_classification_NN_dec.py
349 lines (255 loc) · 13.5 KB
/
thesis_classification_NN_dec.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
#%%
import numpy as np
import tensorflow as tf
from tensorflow.keras import backend as K
from read_root import read_data
from tensorflow import keras
from tensorflow.keras import regularizers
import matplotlib.pyplot as plt
from tensorflow.keras.metrics import Precision, Recall
get_data = read_data()
print("Loading Indices")
path_root_0mm = "/net/data_g4rt/projects/SiFiCC/InputforNN/SiPMNNNewGeometry/FinalDetectorVersion_RasterCoupling_OPM_38e8protons.root"
df_id = get_data.get_df_from_root(path=path_root_0mm, root_entry_str='SiPMData.fSiPMId')
df_id_count = df_id.count(axis=1)
indices_BP0mm = np.where(df_id_count > 1)[0]
path_root_5mm = "/net/data_g4rt/projects/SiFiCC/InputforNN/SiPMNNNewGeometry/FinalDetectorVersion_RasterCoupling_OPM_BP5mm_4e9protons.root"
df_id = get_data.get_df_from_root(path=path_root_5mm, root_entry_str='SiPMData.fSiPMId')
df_id_count = df_id.count(axis=1)
indices_BP5mm = np.where(df_id_count > 1)[0]
del df_id
del df_id_count
print("Loading Data")
############### loading data ###############################
path = "./thesis_models"
model_name = "/decreasing_3103_3212_norm26k_declr"
model_name = path+model_name
data_path = r"/.automount/home/home__home2/institut_3b/farah/Desktop/Data/"
checkpoint_filepath = model_name+"/best_model_fourthmodel_shuffeldBP_step001_layernorm.h5"
#data_path_1232_12k = r"/disk1/farah/"
input_data_BP0mm = np.load(data_path+"training_data_bothneg_norm26k_1632_2ch_midempty_0mmBP.npz")
input_data_BP5mm = np.load(data_path+"training_data_bothneg_norm26k_1632_2ch_midempty_5mmBP.npz")
#input_data_BP0mm = np.load(data_path_1232_12k+"training_data_bothneg_norm12k_1232_2ch_midempty_0mmBP.npz")
#input_data_BP5mm = np.load(data_path_1232_12k+"training_data_bothneg_norm12k_1232_2ch_midempty_5mmBP.npz")
output_data_BP0mm = np.load(data_path+"ideal_targets_BP0mm_ep.npz")
output_data_BP5mm = np.load(data_path+"ideal_targets_BP5mm_ep.npz")
sample_weights_BP0mm = np.load(data_path+"sample_weight_ideal_class_BP0mm_ep.npz")
sample_weights_BP5mm = np.load(data_path+"sample_weight_ideal_class_BP5mm_ep.npz")
#new_indices = new_indices['arr_0']
input_data_BP0mm = input_data_BP0mm['arr_0']
input_data_BP0mm = input_data_BP0mm[indices_BP0mm]
input_data_BP5mm = input_data_BP5mm['arr_0']
input_data_BP5mm = input_data_BP5mm[indices_BP5mm]
output_data_BP0mm = output_data_BP0mm['arr_0']
output_data_BP0mm = output_data_BP0mm[indices_BP0mm]
output_data_BP5mm = output_data_BP5mm['arr_0']
output_data_BP5mm = output_data_BP5mm[indices_BP5mm]
sample_weights_BP0mm = sample_weights_BP0mm['arr_0']
sample_weights_BP0mm = sample_weights_BP0mm[indices_BP0mm]
sample_weights_BP5mm = sample_weights_BP5mm['arr_0']
sample_weights_BP5mm = sample_weights_BP5mm[indices_BP5mm]
print("Slicing Data")
############### slice data ###############################
trainset_index_BP0mm = int(input_data_BP0mm.shape[0]*0.6)
trainset_index_BP5mm = int(input_data_BP5mm.shape[0]*0.6)
valset_index_BP0mm = int(input_data_BP0mm.shape[0]*0.8)
valset_index_BP5mm = int(input_data_BP5mm.shape[0]*0.8)
X_train_BP0mm = input_data_BP0mm[:trainset_index_BP0mm]
X_train_BP5mm = input_data_BP5mm[:trainset_index_BP5mm]
X_val_BP0mm = input_data_BP0mm[valset_index_BP0mm:]
X_val_BP5mm = input_data_BP5mm[valset_index_BP5mm:]
X_test_BP0mm = input_data_BP0mm[trainset_index_BP0mm:valset_index_BP0mm]
X_test_BP5mm = input_data_BP5mm[trainset_index_BP5mm:valset_index_BP5mm]
Y_train_BP0mm = output_data_BP0mm[:trainset_index_BP0mm]
Y_train_BP5mm = output_data_BP5mm[:trainset_index_BP5mm]
Y_val_BP0mm = output_data_BP0mm[valset_index_BP0mm:]
Y_val_BP5mm = output_data_BP5mm[valset_index_BP5mm:]
Y_test_BP0mm = output_data_BP0mm[trainset_index_BP0mm:valset_index_BP0mm]
Y_test_BP5mm = output_data_BP5mm[trainset_index_BP5mm:valset_index_BP5mm]
sample_weights_train_BP0mm = sample_weights_BP0mm[:trainset_index_BP0mm]
sample_weights_train_BP5mm = sample_weights_BP5mm[:trainset_index_BP5mm]
sample_weights_val_BP0mm = sample_weights_BP0mm[valset_index_BP0mm:]
sample_weights_val_BP5mm = sample_weights_BP5mm[valset_index_BP5mm:]
sample_weights_test_BP0mm = sample_weights_BP0mm[trainset_index_BP0mm:valset_index_BP0mm]
sample_weights_test_BP5mm = sample_weights_BP5mm[trainset_index_BP5mm:valset_index_BP5mm]
X_train = np.concatenate([X_train_BP0mm,X_train_BP5mm])
Y_train = np.concatenate([Y_train_BP0mm,Y_train_BP5mm])
sample_weights_train = np.concatenate([sample_weights_train_BP0mm,sample_weights_train_BP5mm])
X_val= np.concatenate([X_val_BP0mm,X_val_BP5mm])
Y_val= np.concatenate([Y_val_BP0mm,Y_val_BP5mm])
sample_weights_val= np.concatenate([sample_weights_val_BP0mm,sample_weights_val_BP5mm])
X_test = np.concatenate([X_test_BP0mm,X_test_BP5mm])
Y_test = np.concatenate([Y_test_BP0mm,Y_test_BP5mm])
sample_weights_test= np.concatenate([sample_weights_test_BP0mm,sample_weights_test_BP5mm])
print("Shuffling Data")
########## shuflle the data #######################
permutation = np.random.permutation(len(X_train))
X_train = X_train[permutation]
Y_train = Y_train[permutation]
sample_weights_train = sample_weights_train[permutation]
permutation = np.random.permutation(len(X_val))
X_val= X_val[permutation]
Y_val= Y_val[permutation]
sample_weights_val= sample_weights_val[permutation]
######################################################
# Define the threshold for computing precision and recall
precision = Precision(thresholds=0.5,name="precision")
recall = Recall(thresholds=0.5,name="recall")
## notes: pool size 2,2,1 is better than 2,2,2
model = keras.Sequential([keras.layers.InputLayer(input_shape = (16,32,2,2)),
keras.layers.Conv3D(filters = 64, kernel_size = (3,3,2),padding="same",activation="relu",kernel_initializer='glorot_normal'),
# keras.layers.Dropout(0.3),
#keras.layers.MaxPool3D(pool_size=(2,1,2),padding="same"),
keras.layers.LayerNormalization(),
keras.layers.Conv3D(filters = 32, kernel_size = (3,3,2),padding="same",activation="relu",kernel_initializer='glorot_normal'),
keras.layers.Dropout(0.3),
keras.layers.MaxPool3D(pool_size=(2,1,2),padding="same"),
keras.layers.LayerNormalization(),
keras.layers.Conv3D(filters = 16, kernel_size = (3,3,2),padding="same",activation="relu",kernel_initializer='glorot_normal'),
keras.layers.Dropout(0.3),
keras.layers.MaxPool3D(pool_size=(2,1,2),padding="same"),
keras.layers.LayerNormalization(),
keras.layers.Conv3D(filters = 8, kernel_size = (3,3,2),padding="same",activation="relu",kernel_initializer='glorot_normal'),#,kernel_initializer='he_uniform'),
keras.layers.Conv3D(filters = 4, kernel_size = (3,3,2),padding="same",activation="relu",kernel_initializer='glorot_normal'),#,kernel_initializer='he_uniform'),
keras.layers.Conv3D(filters = 2, kernel_size = (3,3,2),padding="same",activation="relu",kernel_initializer='glorot_normal'),
keras.layers.Flatten(),
keras.layers.Dense(200, activation = "relu"),
keras.layers.Dropout(0.3),
keras.layers.Dense(50, activation = "relu"),
keras.layers.Dense(1, activation = "sigmoid")])
model.compile(loss='binary_crossentropy',
optimizer = keras.optimizers.Adam(0.001,epsilon=0.0001),
metrics=[precision,recall],
weighted_metrics=['accuracy'] )
# model summary
model.summary()
## print data shape and compton percentage:
print("##################### \n percentage of Compton events in Training set is: ", len(Y_train[Y_train == 1])/len(Y_train), "\n ##################### \n")
print("##################### \n percentage of Compton events in Val set is: ", len(Y_val[Y_val == 1])/len(Y_val), "\n ##################### \n")
print("##################### \n percentage of Compton events in Test set is: ", len(Y_test[Y_test == 1])/len(Y_test), "\n ##################### \n")
print("################### \n train set size is:", X_train.shape, "\n ##################### \n")
print("################### \n val set size is:", X_val.shape, "\n ##################### \n")
print("################### \n complete data set size is:", len(input_data_BP0mm) + len(input_data_BP5mm) , "\n ##################### \n")
#print("################### \n complete data set shape is:", input_data.shape, "\n ##################### \n")
# train model
history = model.fit(X_train,
Y_train,
epochs = 500,
validation_data = (X_val, Y_val,sample_weights_val),
callbacks = [tf.keras.callbacks.ReduceLROnPlateau('val_loss', factor=0.1,patience=10,min_lr=0,verbose=1),
tf.keras.callbacks.EarlyStopping('val_loss', patience=15),
tf.keras.callbacks.ModelCheckpoint(
checkpoint_filepath,
monitor='val_loss',
save_best_only=True,
mode='min',
verbose=1
)],
batch_size = 64,
sample_weight=sample_weights_train)
# class_weight = {0:0.59396336, 1:3.16061141})
# )##tf.keras.callbacks.EarlyStopping('val_loss', patience=30),
## weight for dist compton {0:0.63047041, 1:2.41614324}
## weight for cut at 1, 10 {0:0.62772195, 1:2.45737701})
## weight for ideal {0:0.59396336, 1:3.16061141}
## weight for ideal cut at 2,20 {0:0.57175812 1:3.98392621}
## weight for dist cut at 2,20 {0:0.59540689,1:3.12035585}
#evaluate model
score = model.evaluate(X_test, Y_test, verbose = 1, sample_weight=sample_weights_test)
print('Test loss:', score[0])
print('Test accuracy:', score[1])
y_pred = model.predict(X_test)
model.save(model_name)
#np.savez("y_pred_deep.npz",y_pred)
index_pred = np.where(y_pred > 0.5)
num_entries = len(Y_test[index_pred[0]])
n_compton = Y_test[index_pred[0]].sum()
efficiency = n_compton/len(Y_test[Y_test==1])
Purity = n_compton/len(index_pred[0])
print("Efficiency is", efficiency)
print("Purity is", Purity)
print("##################### \n percentage of Compton events in Predicted set for threshold 0.5 is: ", len(index_pred[0])/len(y_pred), "\n ##################### \n")
index_pred = np.where(y_pred >= 0.6)
num_entries = len(Y_test[index_pred[0]])
n_compton = Y_test[index_pred[0]].sum()
efficiency = n_compton/len(Y_test[Y_test==1])
Purity = n_compton/len(index_pred[0])
print("Efficiency is", efficiency)
print("Purity is", Purity)
print("##################### \n percentage of Compton events in Predicted set for threshold 0.6 is: ", len(index_pred[0])/len(y_pred), "\n ##################### \n")
#%%
index_pred = np.where(y_pred >= 0.8)
num_entries = len(Y_test[index_pred[0]])
n_compton = Y_test[index_pred[0]].sum()
efficiency = n_compton/len(Y_test[Y_test==1])
Purity = n_compton/len(index_pred[0])
print("Efficiency is", efficiency)
print("Purity is", Purity)
print("##################### \n percentage of Compton events in Predicted set for threshold 0.8 is: ", len(index_pred[0])/len(y_pred), "\n ##################### \n")
#%%
np.savetxt(model_name+"_loss.csv", history.history['loss'], delimiter=',')
np.savetxt(model_name+"_val_loss.csv", history.history['val_loss'], delimiter=',')
np.savetxt(model_name+"_acc.csv", history.history['accuracy'], delimiter=',')
np.savetxt(model_name+"_val_acc.csv", history.history['val_accuracy'], delimiter=',')
np.savetxt(model_name+"_precision.csv", history.history['precision'], delimiter=',')
np.savetxt(model_name+"_val_precision.csv", history.history['val_precision'], delimiter=',')
#np.savetxt(model_name+"_precision_08.csv", history.history['precision_08'], delimiter=',')
#np.savetxt(model_name+"_val_precision_08.csv", history.history['val_precision_08'], delimiter=',')
np.savetxt(model_name+"_recall.csv", history.history['recall'], delimiter=',')
np.savetxt(model_name+"_val_recall.csv", history.history['val_recall'], delimiter=',')
#np.savetxt(model_name+"_recall_08.csv", history.history['recall_08'], delimiter=',')
#np.savetxt(model_name+"_val_recall_08.csv", history.history['val_recall_08'], delimiter=',')
plt.rcParams["font.family"] = "serif"
#%%
# summarize history for loss
plt.figure(0)
plt.plot(history.history['loss'],c="blue")
plt.plot(history.history['val_loss'],c="green")
plt.grid(True)
plt.title('model loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'val'], loc='upper left')
plt.savefig(model_name+'_loss_hist1.png')
# summarize history for accuracy
plt.figure(2)
plt.plot(history.history['accuracy'],label="balanced acc",c="k")
plt.plot(history.history['val_accuracy'],label="val balanced acc",c="red")
plt.grid(True)
plt.title('model balanced accuracy')
plt.ylabel('acc')
plt.xlabel('epoch')
plt.legend()
plt.savefig(model_name+'_balanced_acc_hist.png')
#%%
plt.figure(3)
plt.plot(history.history['precision'],label="train purity",c="k")
plt.plot(history.history['val_precision'],label="val purity",c="red")
plt.title('model purity scores threshold 0.5')
plt.ylabel('purity')
plt.grid(True)
plt.xlabel('epoch')
plt.legend()
plt.savefig(model_name+'_precision.png')
#%%
plt.figure(5)
plt.plot(history.history['recall'],label="train effieciency",c="blue")
plt.plot(history.history['val_recall'],label="val effieciency",c="green")
plt.grid(True)
plt.title('model ')
plt.title('model efficiency scores threshold 0.5')
plt.xlabel('epoch')
plt.legend()
plt.savefig(model_name+'_recall.png')
#%%
plt.figure(8)
plt.plot(history.history['precision'],label="train precision",c="k")
plt.plot(history.history['val_precision'],label="val precision",c="red")
plt.plot(history.history['recall'],label="train efficiency",c="blue")
plt.plot(history.history['val_recall'],label="val efficiency",c="green")
plt.grid(True)
plt.title('model purity and efficiency scores threshold 0.5')
plt.ylabel('scores')
plt.xlabel('epoch')
plt.legend()
plt.savefig(model_name+'_val_precision_recall_05.png')