def main():
train_data, train_labels, test_data, test_labels = data.load_all_data_from_zip('hw5digits.zip','hw5')
means = compute_mean_mles(train_data, train_labels)
covariances = compute_sigma_mles(train_data, train_labels, means)
print("############## For train_data #############")
predicted_train_labels = classify_data(train_data, means, covariances)
print("Accuracy = ", accuracy_score(train_labels, predicted_train_labels))
print("Average conditional likelihood = ", avg_conditional_likelihood(train_data, train_labels, means, covariances))
print("############## For test_data #############")
predicted_test_labels = classify_data(test_data, means, covariances)
print("Accuracy = ", accuracy_score(test_labels, predicted_test_labels))
print("Average conditional likelihood = ", avg_conditional_likelihood(test_data, test_labels, means, covariances))
fig=plt.figure(figsize=(8, 8))
columns = 2
rows = 5
for i in range(0,10):
cov = covariances[i,:,:]
w, v = np.linalg.eig(cov)
eig_vals_sorted = np.sort(w)
eig_vecs_sorted = v[:, w.argsort()]
#Computing the leading eigenvectors (largest eigenvalue) for each class covariance matrix
temp_image = eig_vecs_sorted[:,63].reshape(8,8)
#print("temp_image:", temp_image)
fig.add_subplot(rows, columns, i+1)
plt.imshow(temp_image)
plt.show()
def svmmain() -> svm.LinearSVC:
train_data, train_labels, test_data, test_labels = data.load_all_data_from_zip(
'a2digits.zip', 'data')
clf = svm.LinearSVC()
clf.fit(train_data, train_labels)
print("accuracy for SVM on test set is ",
(clf.predict(test_data) == test_labels).sum() / 4000)
return clf
def process_data():
train_data, train_labels, test_data, test_labels = data.load_all_data_from_zip(
'a3digits.zip', 'data')
train_labels, test_labels = label_to_oneht(train_labels), label_to_oneht(
test_labels)
train_data, test_data = torch.from_numpy(train_data), torch.from_numpy(
test_data)
train_data, train_labels, test_data, test_labels = Variable(train_data), Variable(train_labels) \
, Variable(test_data), Variable(test_labels)
return train_data, train_labels, test_data, test_labels
def main():
train_data, train_labels, test_data, test_labels = \
data.load_all_data_from_zip('hw4digits.zip', 'data')
# Fit the model
means = compute_mean_mles(train_data, train_labels)
covariances = compute_sigma_mles(train_data, train_labels)
# Evaluation
# 2(a)
train_avg_cond = \
avg_conditional_likelihood(train_data, train_labels, means, covariances)
test_avg_cond = \
avg_conditional_likelihood(test_data, test_labels, means, covariances)
print("The avg_conditional_likelihood for TRAIN: \n "
"{}".format(train_avg_cond))
print("The avg_conditional_likelihood for TEST: \n "
"{}".format(test_avg_cond))
# 2(b) find predictions of training data and test data,
# and find the accurarcy
train_predictions = classify_data(train_data, means, covariances)
test_predictions = classify_data(test_data, means, covariances)
num_correct_train = np.count_nonzero(train_predictions - train_labels == 0)
num_correct_test = np.count_nonzero(test_predictions - test_labels == 0)
print("train accuracy: ", num_correct_train / len(train_data))
print("test accuracy: ", num_correct_test / len(test_data))
# 2(c) Compute leading eigenvectors for each class cov matrix
# plot them side by side as 8 by 8 images
for k in range(10):
w, v = np.linalg.eig(covariances[k])
max_w_index = np.where(w == max(w))
leading_v = v[:, max_w_index]
# print("The leading vector for k={} is: \n {}".format(k, leading_v))
plt.figure()
im = plt.imshow(leading_v.reshape(8, 8), cmap='gray')
plt.savefig("{}.png".format(k))
plt.show()
Here you should load the data and plot
the means for each of the digit classes.
'''
import data
import numpy as np
# Import pyplot - plt.imshow is useful!
import matplotlib.pyplot as plt
def plot_means(train_data, train_labels):
means = []
for i in range(0, 10):
i_digits = data.get_digits_by_label(train_data, train_labels, i)
# Compute mean of class i
sum_digit = np.sum(i_digits, axis=0)
mean_vecotr = sum_digit / len(i_digits)
mean = mean_vecotr.reshape((8, 8))
means.append(mean)
# Plot all means on same axis
all_concat = np.concatenate(means, 1)
plt.imshow(all_concat, cmap='gray')
plt.show()
if __name__ == '__main__':
train_data, train_labels, _, _ = data.load_all_data_from_zip(
'a3digits.zip', 'data')
plot_means(train_data, train_labels)
import data
import numpy as np
# Import pyplot - plt.imshow is useful!
import matplotlib.pyplot as plt
def plot_means(train_data, train_labels):
means = []
for i in range(0, 10):
i_digits = data.get_digits_by_label(train_data, train_labels, i)
# Compute mean of class i
column = []
column = np.mean(i_digits, axis=0)
means.append(np.reshape(column, (8, 8)))
# Plot all means on same axis
all_concat = np.concatenate(means, 1)
plt.imshow(all_concat, cmap='gray')
# for i in range(len(means)):
# plt.subplot(1,10,i+1)
# plt.imshow(means[i], cmap='gray')
# plt.suptitle('Mean for digit 0-9')
plt.show()
if __name__ == '__main__':
train_data, train_labels, test_data, teste_labels = data.load_all_data_from_zip(
'a2digits.zip', 'data')
plot_means(train_data, train_labels)
def main():
train_data, train_labels, _, _ = data.load_all_data_from_zip(
'a2digits.zip', 'data')
plot_means(train_data, train_labels)
def main():
train_data, _, _, _ = data.load_all_data_from_zip('a2digits.zip',
'data',
shuffle=False)
plot_means(train_data)
'''
Question 2.0 Skeleton Code
Here you should load the data and plot
the means for each of the digit classes.
'''
import data
import numpy as np
# Import pyplot - plt.imshow is useful!
import matplotlib.pyplot as plt
def plot_means(train_data, train_labels):
means = []
for i in range(0, 10):
i_digits = data.get_digits_by_label(train_data, train_labels, i)
# Compute mean of class i
means.append(np.mean(i_digits, axis=0).reshape((8,8)))
# Plot all means on same axis
all_concat = np.concatenate(means, 1)
plt.title("Means of Pixel Values for each Digit Class")
plt.imshow(all_concat, cmap='gray')
plt.show()
if __name__ == '__main__':
train_data, train_labels, _, _ = data.load_all_data_from_zip('a2digits.zip', 'data')
plot_means(train_data, train_labels)
