plt.xlabel('Predicted label')
def write_log(callback, names, logs, batch_no):
for name, value in zip(names, logs):
summary = tf.Summary()
summary_value = summary.value.add()
summary_value.simple_value = value
summary_value.tag = name
callback.writer.add_summary(summary, batch_no)
callback.writer.flush()
matplotlib.use("Qt5Agg")
data = dataSet()
train_data, train_alert_labels, train_rul_minutes, train_condition, test_data, test_alert_labels, test_rul_minutes, test_condition = data.get_all_cache_data(
)
train_rul_minutes = train_rul_minutes.reshape(-1, 1)
test_rul_minutes = test_rul_minutes.reshape(-1, 1)
PREDICT = True
TOTAL_EPOCH = 10000
SAVE_MODEL_PER_EPOCH = 50
BATCH_SIZE = 64
# 238
INITIAL_EPOCH = 79
TOTAL_BATCH_NUM = train_alert_labels.shape[0]
INPUT_SIZE = 64
OUTPUT_SIZE = 10
# check if GPU is feasible for training
train_on_gpu = torch.cuda.is_available()
if not train_on_gpu:
print('Training on CPU ...')
else:
print('Training on GPU ...')
criterionMSE = nn.MSELoss()
criterionCE = nn.CrossEntropyLoss()
train_set = dataSet('data\optdigits.tra', 8)
train_loader = torch.utils.data.DataLoader(train_set,
batch_size=32,
shuffle=True)
test_set = dataSet('data\optdigits.tes', 8)
test_loader = torch.utils.data.DataLoader(test_set,
batch_size=32,
shuffle=True)
def testing(model, criterion, load_path, data_loader, gpu, CE=True):
'''
get accuracy of the model
'''
# load the model
model.load_state_dict(torch.load(load_path))
flatten_layer = NetTool.create_flatten_layer(inception9_concat)
fc_input_size = flatten_layer.get_shape()[1:4].num_elements()
fc_layer = NetTool.create_fc_layer(flatten_layer, [fc_input_size, 1000], 0.4)
out_layer = NetTool.create_fc_layer(fc_layer, [1000, 3], 0.4, use_relu=False)
pred_Y = tf.nn.softmax(out_layer, name='pred_Y')
loss = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(labels=Y, logits=pred_Y))
optimizer = tf.train.AdamOptimizer().minimize(loss) # learning_rate=0.0001
temp = tf.equal(tf.arg_max(pred_Y, 1), tf.arg_max(Y, 1))
accuracy = tf.reduce_mean(tf.cast(temp, tf.float32))
print('开始加载训练数据集')
trainSet = dataset.dataSet(filePath, classes, way='txt', txtPath=txtPath)
print('开始加载测试数据集')
txtFilePath = '/Volumes/Seagate Backup Plus Drive/服务外包/picture/2019-03-05/body1'
testSet = dataset.dataSet(txtFilePath, classes, way='image')
print('数据集加载完成')
saver = tf.train.Saver()
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
for i in range(10001):
batchX, batchY, _ = trainSet.next_batch(32)
sess.run(optimizer, feed_dict={X: batchX, Y: batchY})
if i % 25 == 0:
_, train_ac = sess.run([optimizer, accuracy],
feed_dict={
def __init__(self, rootSet):
self.rootNode = Node(rootSet.getLabel())
self.rootSet = rootSet
def printTree(self):
for pre, fill, node in RenderTree(self.rootNode):
print('%s%s' % (pre, node.name))
def generateBranches(self, currentDataSet, parentNode):
if currentDataSet.isLeaf():
# print('Is leaf', currentDataSet.getTuples()[0])
#class variable
classVar = currentDataSet.getTuples()[0][0]
Node(currentDataSet.getClassName() + '='+classVar, parent=parentNode)
# Start splitting tree and recursing
else:
splitting_attribute_index = currentDataSet.getSplitAttributeByGain()['attribute_index']
split_data = currentDataSet.splitDataOnAttribute(splitting_attribute_index)
for branch in split_data:
nodeLabel = branch.getLabel()
print('node label',nodeLabel)
newNode = Node(nodeLabel, parent=parentNode)
self.generateBranches(branch, newNode)
def runTree(self):
self.generateBranches(self.rootSet, self.rootNode)
self.printTree()
records = CSVReader('../Sources/Wine/red.csv').read([1,2,3,4,5,6,7])
# print(records)
recordSet = dataSet(records['data'], records['attributes'], 'root')
testTree = decisionTree(recordSet)
testTree.runTree()