/
Net_Collision_Classification.py
163 lines (142 loc) · 7.6 KB
/
Net_Collision_Classification.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
import tensorflow as tf
from tensorflow.keras.models import Sequential, Model
from tensorflow.keras.layers import Dense, Flatten, LeakyReLU, Dropout, PReLU, SimpleRNN, LSTM
import scipy as sp
import matplotlib.pyplot as plt
from tensorflow.keras.models import load_model
import CalculateConserved as CalcConsd
import time
start_time = time.time()
# Predicts whether an initial state leads to a collision or not.
# Author: Lawson Fuller
#===============================================================================================================
# Run Parameters
#===============================================================================================================
d_train = sp.load("LabValuesTrain_Misses.npz")
d_test = sp.load("LabValuesIntermediate_Misses.npz")
plotHistory = True
numSamples = 10000
T = 199 # max number of timesteps in matrix
batchSize = 32
numEpochs = 100
#===============================================================================================================
# Import Simulation Data and Preprocess
#===============================================================================================================
Position1L_t= sp.transpose(d_train['arr_11'][:,:numSamples,:], (1,2,0)) # (samples, timsteps, features)
Position2L_t= sp.transpose(d_train['arr_12'][:,:numSamples,:], (1,2,0))
Velocity1L_t= sp.transpose(d_train['arr_13'][:,:numSamples,:], (1,2,0))
Velocity2L_t= sp.transpose(d_train['arr_14'][:,:numSamples,:], (1,2,0))
isCollision_Arr = d_train['arr_18'] [:numSamples]
radius_Arr = sp.array([d_train['arr_19'] [:numSamples]] ).transpose()
# isCollision_Arr= sp.delete( sp.repeat(d_test['arr_18'][:numSamples,sp.newaxis],sp.shape(Velocity1L_t)[1],axis=1),
# [t for t in range(1,sp.shape(Velocity1L_t)[1])],1)
# radius_Arr= sp.delete( sp.repeat(d_test['arr_19'][:numSamples,sp.newaxis],sp.shape(Velocity1L_t)[1],axis=1),
# [t for t in range(1,sp.shape(Velocity1L_t)[1])],1)
Position1L_first= sp.delete(Position1L_t, [t for t in range(1,sp.shape(Position1L_t)[1])],1)
Position2L_first= sp.delete(Position2L_t, [t for t in range(1,sp.shape(Position2L_t)[1])],1)
Velocity1L_first= sp.delete(Velocity1L_t, [t for t in range(1,sp.shape(Velocity1L_t)[1])],1)
Velocity2L_first= sp.delete(Velocity2L_t, [t for t in range(1,sp.shape(Velocity2L_t)[1])],1)
a = sp.array(sp.repeat(d_test['arr_15'][:numSamples,sp.newaxis],T,axis=1))
# m1_Arr= sp.delete( sp.repeat(d_test['arr_15'][:numSamples,sp.newaxis],sp.shape(Velocity1L_t)[1],axis=1),
# [t for t in range(1,sp.shape(Velocity1L_t)[1]-1)],1)
# m2_Arr= sp.delete( sp.repeat(d_test['arr_16'][:numSamples,sp.newaxis],sp.shape(Velocity1L_t)[1],axis=1),
# [t for t in range(1,sp.shape(Velocity1L_t)[1]-1)],1)
# dt_Arr=sp.repeat(d_train['arr_17'][:numSamples,sp.newaxis],T,axis=1)
# E_i=d_train['arr_4'][:numSamples]
# E_f=d_train['arr_5'][:numSamples]
# p_x_i=d_train['arr_6'][:numSamples]
# p_x_f=d_train['arr_7'][:numSamples]
# p_y_i=d_train['arr_8'][:numSamples]
# p_y_f=d_train['arr_9'][:numSamples]
Position1L_t_val= sp.transpose(d_test['arr_11'][:,:numSamples,:], (1,2,0)) # (samples, timesteps, features)
Position2L_t_val= sp.transpose(d_test['arr_12'][:,:numSamples,:], (1,2,0))
Velocity1L_t_val= sp.transpose(d_test['arr_13'][:,:numSamples,:], (1,2,0))
Velocity2L_t_val= sp.transpose(d_test['arr_14'][:,:numSamples,:], (1,2,0))
isCollision_Arr_val = d_test['arr_18'] [:numSamples]
radius_Arr_val = sp.array([d_test['arr_19'] [:numSamples]]).transpose()
Position1L_first_val= sp.delete(Position1L_t_val, [t for t in range(1,sp.shape(Position1L_t_val)[1])],1)
Position2L_first_val= sp.delete(Position2L_t_val, [t for t in range(1,sp.shape(Position2L_t_val)[1])],1)
Velocity1L_first_val= sp.delete(Velocity1L_t_val, [t for t in range(1,sp.shape(Velocity1L_t_val)[1])],1)
Velocity2L_first_val= sp.delete(Velocity2L_t_val, [t for t in range(1,sp.shape(Velocity2L_t_val)[1])],1)
# m1_Arr_val= sp.delete( sp.repeat(d_test['arr_15'][:numSamples,sp.newaxis],sp.shape(Velocity1L_t)[1],axis=1),
# [t for t in range(1,sp.shape(Velocity1L_t_val)[1]-1)],1)
# m2_Arr_val= sp.delete( sp.repeat(d_test['arr_16'][:numSamples,sp.newaxis],sp.shape(Velocity1L_t)[1],axis=1),
# [t for t in range(1,sp.shape(Velocity1L_t_val)[1]-1)],1)
# dt_Arr_val=sp.repeat(d_test['arr_17'][:numSamples,sp.newaxis],T,axis=1)
E_i_val=d_test['arr_4'][:numSamples]
E_f_val=d_test['arr_5'][:numSamples]
p_x_i_val=d_test['arr_6'][:numSamples]
p_x_f_val=d_test['arr_7'][:numSamples]
p_y_i_val=d_test['arr_8'][:numSamples]
p_y_f_val=d_test['arr_9'][:numSamples]
#===============================================================================================================
# Collect Input and Output for Network
#===============================================================================================================
# Dimensions of (samples, timesteps, features)
input_Arr = sp.dstack((Position1L_first, Position2L_first, Velocity1L_first, Velocity2L_first,
radius_Arr))[:,0,:]
target_Arr = isCollision_Arr
input_Arr_val = sp.dstack((Position1L_first_val, Position2L_first_val, Velocity1L_first_val, Velocity2L_first_val,
radius_Arr_val))[:,0,:]
target_Arr_val = isCollision_Arr_val
#===============================================================================================================
# Network
#===============================================================================================================
# Training existing model. Comment out if you do not wish to do this.
model = load_model("trainedModel_temp.hd5")
model = Sequential()
model.add(Dense(sp.shape(input_Arr)[1]*3, input_shape= sp.shape(input_Arr[0]), activation='linear'))
model.add(LeakyReLU(alpha=0.3))
model.add(Dense(sp.shape(input_Arr)[1]*3, activation='linear'))
model.add(LeakyReLU(alpha=0.3))
model.add(Dense(sp.shape(input_Arr)[1]*3, activation='linear'))
# for i in range(50):
# model.add(Dense(sp.shape(input_Arr)[1], activation='linear'))
#
# # model.add(LeakyReLU(alpha=0.3))
model.add(Dense(1, activation='sigmoid'))
#Compile:
opt = tf.keras.optimizers.Adam(lr=1e-3, decay=1e-6)
model.compile(loss= 'binary_crossentropy',
optimizer = opt,
metrics = ['accuracy'])
history = model.fit(input_Arr, target_Arr, batch_size= batchSize, epochs=numEpochs,
validation_data = (input_Arr_val, target_Arr_val),
use_multiprocessing = True)
model.save("NCC-Output\\trainedModel_NCC.hd5")
print("Time to run: ")
print(str(time.time() - start_time)+" seconds")
classes = sp.array([0,1])
pred_from_val_input = model.predict_classes(input_Arr_val)
falseZero = 0
falseOne = 0
trueZero = 0
trueOne = 0
for i in range(len(target_Arr_val)):
if target_Arr_val[i] == 1:
if pred_from_val_input[i] ==1:
trueOne+=1
else:
falseZero += 1
else:
if pred_from_val_input[i] ==1:
falseOne+=1
else:
trueZero += 1
model.summary()
if plotHistory == True:
# Plot training & validation loss values
plt.figure()
plt.plot(history.history['loss'][1:])
plt.plot(history.history['val_loss'][1:])
plt.title('Model loss')
plt.ylabel('Loss')
plt.xlabel('Epoch')
plt.legend(['Training', 'Validation'], loc='upper left')
axes = plt.gca()
# axes.set_ylim(0, 0.1)
plt.savefig("NCC-Output\\NCC-LossConv" + str(numSamples) + "SamplesAnd" + str(numEpochs) + "Epochs.png")
print((trueOne, falseOne))
print((falseZero, trueZero))
# con_mat = tf.math.confusion_matrix(labels=classes, predictions=pred_from_val_input, num_classes=2, dtype=int)
# print(con_mat[0])