-
Notifications
You must be signed in to change notification settings - Fork 0
/
TFQuantum_classification.py
171 lines (109 loc) · 5.15 KB
/
TFQuantum_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
"""
This is a simple classification model to compare the performane of Quantum models
and hybrid Quantum/classical model. Single layer classical dense network is also
included to check the performance.
"""
import numpy as np
import sympy
import matplotlib.pyplot as plt
import os
# For Quantum ML
import cirq
import tensorflow as tf
from tensorflow.keras.callbacks import EarlyStopping
import tensorflow_quantum as tfq
from DataHandler import Data_Handler
class TFQuantum:
"""
This class creates, trains and test all three models for the given data handler object (individual Dataset).
"""
def __init__(self, dataHandler):
self.dataHandler = dataHandler
def createModel(self, model_circuit, model_readout,type = "Quantum"):
if type == "Quantum":
self.model = tf.keras.Sequential([
tf.keras.layers.Input(shape=(), dtype=tf.string),
tfq.layers.PQC(model_circuit, model_readout),
])
elif type == "Hybrid":
input = tf.keras.Input(shape=() ,dtype=tf.dtypes.string)
quatum_output = tfq.layers.PQC(model_circuit, model_readout)(input)
classifier_output = tf.keras.layers.Dense(2, activation=tf.keras.activations.sigmoid)(quatum_output)
self.model = tf.keras.Model(inputs=input, outputs=classifier_output)
elif type == "Single_Dense":
input = tf.keras.Input(shape=((1,2)))
classifier_output = tf.keras.layers.Dense(5, activation=tf.keras.activations.sigmoid)(input)
classifier_output = tf.keras.layers.Dense(2, activation=tf.keras.activations.sigmoid)(classifier_output)
self.model = tf.keras.Model(inputs=input, outputs=classifier_output)
self.model.compile(
loss="binary_crossentropy",
optimizer=tf.keras.optimizers.Adam(learning_rate=0.1),
metrics=["accuracy"])
def train(self, method = "Quantum"):
if method == "Single_Dense":
train_Circuit, train_label = self.dataHandler.get_trainData()
val_Circuit, val_label = self.dataHandler.get_trainData()
else:
train_Circuit, train_label = self.dataHandler.get_valCircuit()
val_Circuit, val_label = self.dataHandler.get_valCircuit()
callBack = EarlyStopping(patience=10, restore_best_weights=True)
model_history = self.model.fit(
train_Circuit, train_label,
batch_size=200,
epochs=20,
verbose=1,
validation_data=(val_Circuit, val_label), callbacks=[callBack])
def predict(self, method ="Quantum"):
if method == "Single_Dense" or method == "Ground_Truth":
test_Circuit, test_label = self.dataHandler.get_testData()
else:
test_Circuit, test_label = self.dataHandler.get_testCircuit()
if method != "Ground_Truth":
results = self.model.predict(test_Circuit)
self.y_pred = np.argmax(results, axis=1)
evaluation_results = self.model.evaluate(test_Circuit, test_label)
print("Type: {}, Loss: {}, Accuracy: {}".format(method, evaluation_results[0], evaluation_results[1]))
else:
self.y_pred = np.argmax(test_label, axis=1)
def plot(self,axis,plotNu, method):
i = int(plotNu%2)
j = int(plotNu/2)
test_Circuit, test_label = self.dataHandler.get_testData()
ix = np.where(self.y_pred == 0)
axis[i, j].scatter(test_Circuit[ix, 0], test_Circuit[ix, 1], c="red", label="0", s=100)
ix = np.where(self.y_pred == 1)
axis[i, j].scatter(test_Circuit[ix, 0], test_Circuit[ix, 1], c="green", label="1", s=100)
axis[i, j].legend()
axis[i, j].set_title(method)
# A simple Quantum circuit using cirq
input_qubits = cirq.GridQubit.rect(1, 2) # 1x2 grid.
model_circuit = cirq.Circuit()
alpha1 = sympy.Symbol('a1')
model_circuit.append(cirq.rx(alpha1)(input_qubits[0]))
alpha2 = sympy.Symbol('a2')
model_circuit.append(cirq.rx(alpha2)(input_qubits[1]))
alpha3 = sympy.Symbol('a3')
model_circuit.append(cirq.XX(input_qubits[1],input_qubits[0])**alpha3)
alpha4 = sympy.Symbol('a4')
model_circuit.append(cirq.H(input_qubits[0])**alpha4)
alpha5 = sympy.Symbol('a5')
model_circuit.append(cirq.H(input_qubits[1])**alpha5)
model_readout = [cirq.X(input_qubits[0]),cirq.X(input_qubits[1])]
print(model_circuit)
DataList = ["HLine","VLine", "Triangle", "Circle"]
methodList = ["Ground_Truth","Quantum", "Hybrid", "Single_Dense"]
resultsFolder = "results/"
if not os.path.exists(resultsFolder):
os.makedirs(resultsFolder)
fig, ax = plt.subplots(2,2,figsize=(16,16))
for data in DataList:
dataObj = Data_Handler(Nu_Data=15000,pattern=data)
TFQobj = TFQuantum(dataHandler=dataObj)
fig, ax = plt.subplots(2, 2, figsize=(16, 16))
for i, method in enumerate(methodList):
if method != "Ground_Truth":
TFQobj.createModel(model_circuit=model_circuit,model_readout=model_readout,type=method)
TFQobj.train(method=method)
TFQobj.predict(method=method)
TFQobj.plot(axis = ax,plotNu=i,method=method)
plt.savefig(resultsFolder+data+".png")