def __net_input(self, X):
return np.dot(X, self.w_[1:]) + self.w_[0]
def __activation(self, X):
return self.__sigmoid(X)
def predict(self, X):
# 1- Calculate the net input W^T * x
z = self.__net_input(X)
# 2- Return the activated values (0 or 1 classes)
return np.where(self.__activation(z) >= 0.5, 1, 0)
pass
reader = CsvReader("./data/Iris.csv")
iris_features, iris_labels = reader.get_iris_data()
ignore_verginica = [
i for i, v in enumerate(iris_labels) if v == 'Iris-virginica'
]
iris_features = [
v for i, v in enumerate(iris_features) if i not in ignore_verginica
]
iris_labels = [
v for i, v in enumerate(iris_labels) if i not in ignore_verginica
]
print(len(iris_features))
print(len(iris_labels))
from data_reader.reader import CsvReader
from util import *
import tensorflow as tf
import matplotlib.pyplot as plt
import math
reader = CsvReader(
"/home/ahani/PycharmProjects/IrisClassification/data/Iris.csv")
iris_features, iris_labels = reader.get_iris_data()
iris_features, iris_labels = shuffle(iris_features, iris_labels)
binarized_labels = binarize_labels(iris_labels)
INPUT_NEURONS = 4
HIDDEN_NEURONS = 200
OUTPUT_NEURONS = 3
NUM_OF_EPOCHS = 1000
x = tf.placeholder(tf.float32, [None, INPUT_NEURONS])
y_target = tf.placeholder(tf.float32, [None, OUTPUT_NEURONS])
input_hidden_weights = tf.Variable(
tf.random_uniform([INPUT_NEURONS, HIDDEN_NEURONS],
-1.0 / math.sqrt(INPUT_NEURONS), 1.0 /
math.sqrt(INPUT_NEURONS))) #Ini from the given network
input_hidden_bias = tf.Variable(tf.ones([HIDDEN_NEURONS])) # The bias in one
hidden_neurons_values = tf.matmul(x, input_hidden_weights) + input_hidden_bias
hidden_activation_result = tf.nn.softmax(hidden_neurons_values)
def __net_input(self, X):
return np.dot(X, self.w_[1:]) + self.w_[0]
def __activation(self, X):
return self.__sigmoid(X)
def predict(self, X):
# 1- Calculate the net input W^T * x
z = self.__net_input(X)
# 2- Return the activated values (0 or 1 classes)
h = self.__activation(z)
return np.where(self.__activation(z) >= 0.5, 1, 0)
reader = CsvReader("./data/titanic/train.csv")
titanic_features, titanic_labels = reader.get_titanic_data()
print(len(titanic_features))
print(len(titanic_labels))
#titanic_labels = to_onehot(titanic_labels)
#titanic_labels = list(map(lambda v: v.index(max(v)), titanic_labels))
train_x, train_y, test_x, test_y = titanic_features[0:712], titanic_labels[
0:712], titanic_features[712:], titanic_labels[712:]
train_x, train_y, test_x, test_y = np.asarray(train_x), np.asarray(
train_y), np.asarray(test_x), np.asarray(test_y)
train_x, means, stds = standardize(train_x)
from data_reader.reader import CsvReader
from util import *
import tensorflow as tf
import matplotlib.pyplot as plt
import math
reader = CsvReader("/home/ahani/PycharmProjects/IrisClassification/data/Iris.csv")
iris_features, iris_labels = reader.get_iris_data()
iris_features, iris_labels = shuffle(iris_features, iris_labels)
binarized_labels = binarize_labels(iris_labels)
INPUT_NEURONS = 4
HIDDEN_NEURONS = 200
OUTPUT_NEURONS = 3
NUM_OF_EPOCHS = 1000
x = tf.placeholder(tf.float32, [None, INPUT_NEURONS])
y_target = tf.placeholder(tf.float32, [None, OUTPUT_NEURONS])
input_hidden_weights = tf.Variable(
tf.random_uniform([INPUT_NEURONS, HIDDEN_NEURONS], -1.0 / math.sqrt(INPUT_NEURONS), 1.0 / math.sqrt(INPUT_NEURONS))) #Ini from the given network
input_hidden_bias = tf.Variable(tf.ones([HIDDEN_NEURONS])) # The bias in one
hidden_neurons_values = tf.matmul(x, input_hidden_weights) + input_hidden_bias
hidden_activation_result = tf.nn.softmax(hidden_neurons_values)
hidden_output_weights = tf.Variable(
tf.random_uniform([HIDDEN_NEURONS, OUTPUT_NEURONS], -1.0 / math.sqrt(HIDDEN_NEURONS), 1.0 / math.sqrt(HIDDEN_NEURONS)))
hidden_output_bias = tf.Variable(tf.ones([OUTPUT_NEURONS])) # The bias in one