plt.show()
'''
# TODO: Test perceptron_predict function, defined in perceptron.py
yPred = perceptron.perceptron_predict(w, XTrain[0])
# TODO: Test perceptron_train function, defined in perceptron.py
w = np.zeros(XTrain.shape[1])
w = perceptron.perceptron_train(w, XTrain, yTrain, 10)
# TODO: Test RBF_kernel function, defined in perceptron.py
K = perceptron.RBF_kernel(XTrain[0:5], XTrain[0:5], 10.0)
# TODO: Test kernel_perceptron_predict function, defined in perceptron.py
a = np.zeros(XTrain.shape[0])
yPred = perceptron.kernel_perceptron_predict(a, XTrain, yTrain, XTrain[0],
10.0)
# TODO: Test kernel_perceptron_train function, defined in perceptron.py
a = np.zeros(XTrain.shape[0])
a0 = perceptron.kernel_perceptron_train(a, XTrain, yTrain, 5, 100)
# TODO: Run experiments outlined in HW4 PDF
w = np.zeros(XTrain.shape[1])
w = perceptron.perceptron_train(w, XTrain, yTrain, 10)
yPred = np.zeros(len(yTest))
for i in range(len(XTest)):
yPred[i] = perceptron.perceptron_predict(w, XTest[i])
np.sum(np.abs(yTest - yPred)) / (2.0 * len(XTest))
for sigma in [0.01, 0.1, 1.0, 10.0, 100.0, 1000.0]:
a = np.zeros(XTrain.shape[0])
# TODO: Test kernel_perceptron_train function, defined in perceptron.py
# TODO: Run experiments outlined in HW4 PDF
#1
# a0 = np.zeros(XTrain.shape[1])
# a1 = pe.perceptron_train(a0,XTrain,yTrain,10)
# error_num = 0
# for i in range(0,XTest.shape[0]):
# y_pre = pe.perceptron_predict(a1,XTest[i])
# y_real = yTest[i]
# if y_real != y_pre:
# error_num += 1
# error_rate = error_num * 1.0 / XTest.shape[0]
# print error_rate
#2
for sigma in (0.01, 0.1, 1, 10, 100, 1000):
a0 = np.zeros(XTrain.shape[0])
a1 = pe.kernel_perceptron_train(a0, XTrain, yTrain, 2, sigma)
error_num = 0
for i in range(0, XTest.shape[0]):
y1 = pe.kernel_perceptron_predict(a1, XTrain, yTrain, XTest[i], sigma)
y2 = yTest[i]
if y1 != y2:
error_num += 1
error_rate = error_num * 1.0 / XTest.shape[0]
print sigma, error_rate
#3
# yhat = perceptron.perceptron_predict(w0, XTest[i])
# if yhat != yTest[i]:
# err += 1
# print float(err)/len(yTest)
#print w0
# TODO: Test RBF_kernel function, defined in perceptron.py
#sigma = 0.1
#print perceptron.RBF_kernel(XTrain[0], XTrain[1], sigma)
# TODO: Test kernel_perceptron_predict function, defined in perceptron.py
# x = XTrain[0]
# sigma = 0.1
# a = np.zeros((n,1))
#print perceptron.kernel_perceptron_predict(a, XTrain, yTrain, x, sigma)
# TODO: Test kernel_perceptron_train function, defined in perceptron.py
a0 = [0 for i in range(0, len(XTrain))]
sigma = 100
a0 = perceptron.kernel_perceptron_train(a0, XTrain, yTrain, 2, sigma)
err = 0
for i in range(0, len(yTest)):
yhat = perceptron.kernel_perceptron_predict(a0, XTrain, yTrain, XTest[i],
sigma)
if yhat != yTest[i]:
err += 1
print float(err) / len(yTest)
# TODO: Run experiments outlined in HW4 PDF
#get the number of features
d = XTrain.shape[1]
n = XTrain.shape[0]
m = XTest.shape[0]
#experiment 1, original perceptron
w0 = np.zeros((d, 1))
w = perceptron.perceptron_train(w0, XTrain, yTrain, 10)
rate1 = perceptron.perceptron_test(w, XTest, yTest)
print(rate1)
#result: error rate: 0.03833
#experiment 2, kernel perceptron
sigmaList = [0.01, 0.1, 1, 10, 100, 1000]
for sigma in sigmaList:
error_case = 0
a0 = np.zeros((n, 1))
a = perceptron.kernel_perceptron_train(a0, XTrain, yTrain, 2, sigma)
for i in range(0, m):
yHat = perceptron.kernel_perceptron_predict(a, XTrain, yTrain,
XTest[i, :], sigma)
if yHat != yTest[i, 0]:
error_case += 1
error_rate = float(error_case) / m
print("The error rate when sigma= ", sigma, " is: ", error_rate)
#result: error rate:[0.49,0.49,0.075,0.05833,0.08833,0.51]
#when sigma=10, we have the lowest test error
#tess = np.tile(weight, (5, 1))
#print tess
alpha = np.zeros(len(XTrain))
#after_w = pp.perceptron_train(weight, XTrain, yTrain, 10)
#count = 0
#for i in range(len(yTest)):
# ypred = pp.perceptron_predict(after_w, XTest[i])
# if ypred != yTest[i]:
# count = count + 1
#print count / float(len(yTest))
#pp.perceptron_train(after_w, XTest, yTest, 1)
#print pp.kernel_perceptron_predict(alpha, XTrain, yTrain, XTrain[0], 2)
alpha_after = pp.kernel_perceptron_train(alpha, XTrain, yTrain, 2, 0.1)
count = 0
for i in range(len(XTest)):
ypred = pp.kernel_perceptron_predict(alpha_after, XTrain, yTrain, XTest[i],
0.1)
if ypred != yTest[i]:
count = count + 1
print count / float(len(yTest))
pp.kernel_perceptron_train(alpha_after, XTest, yTest, 1, 1)
#pp.RBF_kernel(XTrain, XTrain, 2)
# Visualize the image
#idx = 0
#datapoint = XTrain[idx, 1:]
#plt.imshow(datapoint.reshape((28,28), order = 'F'), cmap='gray')
#plt.show()
# TODO: Test perceptron_predict function, defined in perceptron.py
# TODO: Test perceptron_train function, defined in perceptron.py