def main():
train_data, train_label, test_data, test_label = mnist()
learning_rate = 0.1
num_iterations = 2000
w = np.zeros((train_data.shape[0], 1))
b = 0
parameters, gradients, costs = optimizer(train_data, \
train_label, w, b, learning_rate, num_iterations)
w = parameters["w"]
b = parameters["b"]
train_Pred = classify(train_data, w, b)
test_Pred = classify(test_data, w, b)
temp = np.equal(train_Pred, train_label)
temp2 = np.equal(test_Pred, test_label)
trAcc = (np.sum(temp) / train_label.shape[1]) * 100
teAcc = (np.sum(temp2) / test_label.shape[1]) * 100
print("Accuracy for training set is {} %".format(trAcc))
print("Accuracy for testing set is {} %".format(teAcc))
h = np.arange(1, 201)
plt.plot(h, costs)
plt.savefig("Error vs iterations")
def main():
# getting the sbuset dataset from MNIST
train_data, train_label, test_data, test_label = mnist()
# initialize learning rate and num_iterations
learning_rate = 0.1
num_iterations = 2000
# initialize w as array (d,1) and b as a scalar'
w = np.zeros((train_data.shape[0], 1))
b = 0
# learning the weights by using optimize function
parameters, gradients, costs = optimizer(train_data, \
train_label, w, b, learning_rate, num_iterations)
w = parameters["w"]
b = parameters["b"]
# compute the accuracy for training set and testing set
train_Pred = classify(train_data, w, b)
test_Pred = classify(test_data, w, b)
trAcc = 100 - np.mean(np.abs(train_Pred - train_data)) * 100
teAcc = 100 - np.mean(np.abs(test_Pred - test_data)) * 100
print("Accuracy for training set is {} %".format(trAcc))
print("Accuracy for testing set is {} %".format(teAcc))
def main():
# getting the sbuset dataset from MNIST
train_data, train_label, test_data, test_label = mnist()
# initialize learning rate and num_iterations
learning_rate = 0.1
num_iterations = 2000
# initialize w as array (d,1) and b as a scalar
w = np.random.randint(5, size=(train_data.shape[0], 1))
b = 1
# learning the weights by using optimize function
parameters, gradients, costs = optimizer(train_data, \
train_label, w, b, learning_rate, num_iterations)
w = parameters["w"]
b = parameters["b"]
# compute the accuracy for training set and testing set
train_Pred = classify(train_data, w, b)
test_Pred = classify(test_data, w, b)
trAcc = np.sum(train_Pred == train_label) / float(len(train_Pred[0])) * 100
teAcc = np.sum(test_Pred == test_label) / float(len(test_Pred[0])) * 100
print("Accuracy for training set is {} %".format(trAcc))
print("Accuracy for testing set is {} %".format(teAcc))
plt.plot(range(1, len(costs) + 1), costs, label='Cost')
plt.xlabel('Iterations')
plt.ylabel('Cost')
plt.title('Mini Project1:1.3 Logistic Regression-Cost vs Iterations')
plt.legend()
plt.show()
def main():
# getting the sbuset dataset from MNIST
train_data, train_label, test_data, test_label = mnist()
# initialize learning rate and num_iterations
learning_rate = 0.1
num_iterations = 2000
# initialize w as array (d,1) and b as a scalar
w = np.zeros((train_data.shape[0], 1))
b = 0
# learning the weights by using optimize function
parameters, gradients, costs = optimizer(train_data, \
train_label, w, b, learning_rate, num_iterations)
w = parameters["w"]
b = parameters["b"]
# compute the accuracy for training set and testing set
train_Pred = classify(train_data, w, b)
test_Pred = classify(test_data, w, b)
#print(train_label.shape)
#print(np.sum(train_Pred == train_label))
#np.sum((np.abs(train_Pred == train_label))) / train_Pred.shape[1]
#np.sum((np.abs(test_Pred == test_label))) / train_Pred.shape[1]
trAcc = 100 - (np.sum(
(np.abs(train_Pred - train_label))) / train_Pred.shape[1]) * 100
teAcc = 100 - (np.sum(
(np.abs(test_Pred - test_label))) / train_Pred.shape[1]) * 100
print("Accuracy for training set is {} %".format(trAcc))
print("Accuracy for testing set is {} %".format(teAcc))
plt.plot(costs)
plt.xlabel('iteration')
plt.ylabel('error')
plt.savefig('error_plot.png')