import matplotlib.pyplot as plt
from tensorflow import confusion_matrix
import numpy as np
from tensorflow.keras.utils import to_categorical
from tensorflow.keras import models
import scipy.io as scio
from sklearn.metrics import confusion_matrix
import matplotlib.pyplot as plt
import itertools
from tensorflow.keras import backend as K,Model # ????
import os # ????
import tensorflow as tf # ????
# -*- coding: utf-8 -*-
"""
Created on Thu Jul 30 16:47:47 2020
@author: Noah
"""
import os # ????
import tensorflow as tf # ????
from keras import backend as K, Input, Model # ????
os.environ['CUDA_VISIBLE_DEVICES'] = '0' # ????
config = tf.ConfigProto() # ????
config.gpu_options.allow_growth = True # ????
K.set_session(tf.Session(config=config)) # ????
from keras import models
from keras.layers import Flatten, Dense, BatchNormalization, Dropout, Conv1D, SeparableConv1D, Lambda, Concatenate,
GlobalAveragePooling1D, Activation
input_shape = (128, 2)
inputs = Input(shape=input_shape)
stride = 1
def _group_conv(x, filters, kernel, stride, groups):
channel_axis = 1 if K.image_data_format() == 'channels_first' else -1
in_channels = K.int_shape(x)[channel_axis]
# number of input channels per group
nb_ig = in_channels // groups
# number of output channels per group
nb_og = filters // groups
gc_list = []
# Determine whether the number of filters is divisible by the number of groups
assert filters % groups == 0
for i in range(groups):
if channel_axis == -1:
x_group = Lambda(lambda z: z[:, :, i * nb_ig: (i + 1) * nb_ig])(x)
else:
x_group = Lambda(lambda z: z[:, i * nb_ig: (i + 1) * nb_ig, :])(x)
gc_list.append(Conv1D(filters=nb_og, kernel_size=kernel, strides=stride,
padding='same', use_bias=False)(x_group))
return Concatenate(axis=channel_axis)(gc_list)
def LightNet():
L = 128 #sample points
model = models.Sequential()
x1 = Conv1D(128, 16, activation='relu', padding='same',input_shape=[L,2])(inputs)
x2 = BatchNormalization()(x1)
x3 = Dropout(0.003)(x2)
x4 = _group_conv(x3, filters=64, kernel=8, stride=1, groups=8)
x5 = BatchNormalization()(x4)
x6 = Dropout(0.003)(x5)
x7 = GlobalAveragePooling1D()(x6)
x8 = Dense(9)(x7)
predicts = Activation('softmax')(x8)
model = Model(inputs, predicts)
model.compile(loss='categorical_crossentropy',optimizer='adam', metrics=['accuracy'])
model.summary()
return model
model = LightNet()
model.save('m.hdf5')#看模型的大小
from numpy import array
Data_path = "../10modulations_L=128_train6000_val1000_test10000/"
model = LightNet()
model.load_weights("GlobalModel.hdf5")
data_path = Data_path + "test/snr=" + str(20) + ".mat"
data = scio.loadmat(data_path)
x = data.get('IQ')
N = 10000
y1 = np.zeros([N, 1])
y2 = np.ones([N, 1])
y3 = np.ones([N, 1]) * 2
y4 = np.ones([N, 1]) * 3
y5 = np.ones([N, 1]) * 4
y6 = np.ones([N, 1]) * 5
y7 = np.ones([N, 1]) * 6
y8 = np.ones([N, 1]) * 7
y9 = np.ones([N, 1]) * 8
y_flag = np.vstack((y1, y2, y3, y4, y5, y6, y7, y8, y9))
y_flag = array(y_flag)
y = to_categorical(y_flag)
X_test = x
Y_test = y
[loss, acc] = model.evaluate(X_test, Y_test, batch_size=100, verbose=0)
X_pred = model.predict(X_test)
def plot_confusion_matrix(cm, classes,
normalize=False,
cmap=plt.cm.Blues):
cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
thresh = cm.max() / 2.
for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
# plt.text(j, i, ("%.3f" % cm[i, j]),
plt.text(j, i, ("%.2f" % (cm[i, j])), size=10,
horizontalalignment="center",
color="white" if cm[i, j] > thresh else "black")
plt.imshow(cm, interpolation='nearest', cmap=cmap)
#cbar = plt.colorbar()
#cbar.set_ticks(np.linspace(0, 1, 11))
#cbar.set_ticklabels(('0%', '10%', '20%', '30%', '40%', '50%', '60%', '70%', '80%', '90%', '100%'))
tick_marks = np.arange(len(classes))
plt.xticks(tick_marks, classes, size=14, family='Times New Roman',rotation=30)
plt.yticks(tick_marks, classes, size=14, family='Times New Roman')
plt.title("Confusion Matrix (SNR=20db)", size=14) # 图像标题
plt.ylabel('True label', size=13, family='Times New Roman')
plt.xlabel('Predicted label', size=13, family='Times New Roman')
plt.colorbar()
#confusion_matrix_title = 'Confusion matrix'
# print(confusion_matrix_title)
proba = model.predict(X_test, batch_size=100, verbose=1)
max = np.max(proba, axis=1)
y_pred = np.zeros(shape=(90000, 1))
for i in range(90000):
for j in range(9):
if (proba[i][j]==max[i]):
y_pred[i]=j
cm = confusion_matrix(y_flag, y_pred)
classes = ["2FSK", "4FSK", "8FSK", "BPSK", "QPSK", "8PSK", "16QAM","128QAM", "256QAM"]
plt.figure()
plot_confusion_matrix(cm, classes, normalize=True, cmap=plt.cm.Blues)
plt.show()
data_path = Data_path + "test/snr=" + str(10) + ".mat"
data = scio.loadmat(data_path)
X_test = data.get('IQ')
proba = model.predict(X_test, batch_size=100, verbose=1)
max = np.max(proba, axis=1)
y_pred = np.zeros(shape=(90000, 1))
for i in range(90000):
for j in range(9):
if (proba[i][j] == max[i]):
y_pred[i] = j
cm = confusion_matrix(y_flag, y_pred)
classes = ["2FSK", "4FSK", "8FSK", "BPSK", "QPSK", "8PSK", "16QAM","128QAM", "256QAM"]
plt.figure()
plot_confusion_matrix(cm, classes, normalize=True, cmap=plt.cm.Blues)
plt.savefig("demo9_class_10db.png")
plt.show()
