def main_menu():
dp.clear()
#get information from the user
print("Hello,\n")
print(
"Please insert known information on the attack:\n [property1=numVal1 property2=... e.g. nkill=2 nwound=3] "
)
info = input()
infoarr = info.split(" ")
cols = []
infoVec = []
for inf in infoarr:
tup = inf.split("=")
cols.append(tup[0])
infoVec.append(tup[1])
print("Please choose the property predict :[e.g. weaptype1, attacktype1]")
pred = input("Property: ")
print("Please insert number of categories of this property:")
n_pred = input("num of categories: ")
mdl.SoftMax(infoVec, cols, pred, int(n_pred))
return
def SoftMax(infoVec, cols, pred, n_pred):
#get the data
train_x, train_y, test_x, test_y, SoftMaxSet = dp.getTrainAndTestSoftMax(
cols, pred, n_pred, 'globalterrorismdb_dist.csv')
SoftMaxSet.sort()
print(SoftMaxSet)
#init nub of features and num of categorites
numOfFeatures = len(train_x[0])
numOfCategories = len(train_y[0])
print("Num Of Features ", numOfFeatures, " Num Of Categories ",
numOfCategories)
#input layer
x = tf.placeholder(tf.float32, [None, numOfFeatures])
y_ = tf.placeholder(tf.float32, [None, numOfCategories])
#Wegiths and bias
W = tf.Variable(tf.zeros([numOfFeatures, numOfCategories]))
b = tf.Variable(tf.zeros([numOfCategories]))
#output layer
y = tf.nn.softmax(tf.matmul(x, W) + b)
#loss and update functions
loss = -tf.reduce_mean(y_ * tf.log(y))
update = tf.train.GradientDescentOptimizer(0.000001).minimize(loss)
#TRAINING
print(">>>START TRAINING<<<<")
sess = tf.Session()
sess.run(tf.global_variables_initializer())
axis_y = []
axis_x = []
#correct = tf.equal(findMaxIndex(y),findMaxIndex(y_))
#accuracy = tf.reduce_mean(tf.cast(correct, tf.float32))
for i in range(0, 1000):
__, currentLoss = sess.run([update, loss],
feed_dict={
x: train_x,
y_: train_y
})
if (i % 10 == 0):
#print("current loss at epoch ",i,": ",currentLoss)
axis_y.append(currentLoss)
axis_x.append(i)
plt.ylabel('LOSS')
plt.xlabel('EPHOCH')
plt.title('Train Loss Session')
plt.plot(axis_x, axis_y, 'b--')
print(">>>DONE TRAINING<<<<")
#TEST
print("@@@START TEST@@@")
predictionList = y.eval(session=sess, feed_dict={x: test_x})
total = len(predictionList)
success = 0
#print("len(predictionList) ",len(predictionList)," len(test_y) ",len(test_y))
for k, vec in enumerate(predictionList):
index = findMaxIndex(test_y[k])
predIndex = findMaxIndex(vec)
if (predIndex == index):
success += 1
#print("----SUCCESS-----\nPrediction ", predictionList[k])
#print("Original ", test_y[k])
print("@@@DONE TEST@@@")
print("With Test Accuracy of ", int(success / total * 100.0), "%")
feed = [infoVec]
predict = y.eval(session=sess, feed_dict={x: feed})
for res in predict:
for i, val in enumerate(res):
print(" Probabilty for ", findValuForIndex(i + 1, SoftMaxSet),
pred, " is ", val)
plt.show()
import tensorflow as tf
import numpy as np
import DataPreper as dp
def f(z):
return 1 / (1.0 + tf.exp(-z))
train_data = dp.getTrainNN()
train_x = train_data[0]
train_y = train_data[1]
test_data = dp.getTestNN()
test_x = test_data[0]
test_y = test_data[1]
best = 0
bestS = 0
features = 6
i = 100
while i < 10000:
(layer1, layer2, layer3) = (i, 1, 25)
x = tf.placeholder(tf.float32, [None, features])
y_ = tf.placeholder(tf.float32, [None, 1])
#Hidden Layer 1; size 100 ; input layer
W1 = tf.Variable(tf.truncated_normal([features, layer1], stddev=0.1))
b1 = tf.Variable(tf.constant(0.1, shape=[layer1]))
z1 = tf.add(tf.matmul(x, W1), b1)
a1 = tf.nn.relu(z1)
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
import csv
import DataPreper as dp
def h(x, w, b):
return 1 / (1 + np.exp(-(np.dot(x, w) + b)))
train_x = dp.getTrain()[0]
train_y = dp.getTrain()[1]
test_x = dp.getTest()[0]
test_y = dp.getTest()[1]
w = np.array([0., 0, 0, 0, 0, 0])
b = 0
alpha = 0.001
for iteration in range(1000):
gradient_b = np.mean(1 * (train_y - (h(train_x, w, b))))
gradient_w = np.dot(
(train_y - h(train_x, w, b)), train_x) * 1 / len(train_y)
b += alpha * gradient_b
w += alpha * gradient_w
total = 0
fneg = 0
fpos = 0
import GetData
import DataPreper
import SplitData
import TrainData
import TestData
p = getparams()
np.random.seed(0) # Seed the random number generator
# p["Data"]["LoadFromCache"] = True
# p["Kmean"]["LoadFromCache"] = True
# p["DataProcess"]["LoadFromCache"] = True
folderList = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
images, labels = GetData.get_data(p['Data'], folderList)
image_features = DataPreper.data_prepare(p['DataProcess'], images)
train_x_array, train_y_array, test_x_array, test_y_array = SplitData.split_data_train_test(
p['Split'], image_features)
linear_svms = []
poly_svms = []
decisions_array = []
decisions_poly_array = []
for i in range(-5, 15):
p["Train"]["C_Value"] = np.float_power(10, i)
linear_svm = TrainData.train_data_linear(p['Train'], train_x_array,
train_y_array)
poly_svms = TrainData.train_data_non_linear(p['Train'], train_x_array,
train_y_array)
decisions = TestData.test_linear_svm(p['Test'], linear_svm, test_x_array,
import TestData
import Report
from sklearn.metrics import accuracy_score, confusion_matrix
p = getparams()
data_path = os.path.join("D:\\", "Studies", "Learn", "101_ObjectCategories")
np.random.seed(0) # Seed the random number generator
p["Data"]["BaseDataPath"] = data_path
class_indices = [10, 11, 12, 13, 14, 15, 16, 17, 18, 19]
##### train hyper params
if p['TrainHyper']:
class_indices = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
images, labels = GetData.get_data(p['Data'], class_indices)
data_prepare_options = DataPreper.get_data_grid(p['DataProcess'])
number_of_trains = len(data_prepare_options) * (
len(p['Train']['C_Values']) +
len(p['Train']['C_Values']) * len(p['Train']['Poly_Values']))
linear_svms = [None] * number_of_trains
poly_svms = [None] * number_of_trains
data = []
tests_array = [None] * number_of_trains
tests_dec = [None] * number_of_trains
acc_array = [None] * number_of_trains
conf_array = [None] * number_of_trains
decisions_array = [None] * number_of_trains
decisions_poly_array = [None] * number_of_trains
index = 0
#!/usr/bin/python
import numpy as np
import DataPreper as dp
import tensorflow as tf
from fractions import _gcd
from tensorflow.contrib import rnn
from tensorflow.python.ops import rnn, rnn_cell
train_x,train_y,test_x,test_y = dp.TrainAndTestRNN()
# number of features
n_featurs = train_x[0].size
# data size
total_size = train_y.size
batch_size = _gcd(train_y.size,test_y.size)
# gets the data after the split
# the X data contains stock prices of 19 consecutive days
# the y data is the stock price of the 20th day
# reshape to a 3-dimensional tensor
#train_x = train_x.reshape(train_x.shape + tuple([1]))
#test_x = test_x.reshape(test_x.shape + tuple([1]))
#train_y = train_y.reshape(train_y.shape + tuple([1]))
#test_y = test_y.reshape(test_y.shape + tuple([1]))
# rnn configuration