def run_k_fold_cross_validation(X_train, y_train, num_folds, k, k_accuracy):
X_train_folds, y_train_folds = generate_folds(X_train, y_train, num_folds)
accuracy = 0.0
accuracy_list = []
for i in range(num_folds):
val_fold_x = X_train_folds[i]
val_fold_y = y_train_folds[i]
temp_X_train = np.concatenate(X_train_folds[:i] +
X_train_folds[i + 1:])
temp_y_train = np.concatenate(y_train_folds[:i] +
y_train_folds[i + 1:])
classifier = KNearestNeighbor()
classifier.train(temp_X_train, temp_y_train)
dists = classifier.compute_distances_no_loops(val_fold_x)
val_pred_y = classifier.predict_labels(dists, k)
num_correct = np.sum(val_pred_y == val_fold_y)
accuracy_list.append((float(num_correct) / val_pred_y.shape[0]))
accuracy = accuracy + (float(num_correct) / val_pred_y.shape[0])
k_accuracy[k] = accuracy_list
accuracy = accuracy / num_folds
return accuracy
# two matrices are similar; one of the simplest is the Frobenius norm. In case
# you haven't seen it before, the Frobenius norm of two matrices is the square
# root of the squared sum of differences of all elements; in other words, reshape
# the matrices into vectors and compute the Euclidean distance between them.
print("test the difference between two loop and one loop........")
difference = np.linalg.norm(dists - dists_one, ord='fro')
print('Difference was: %f' % (difference, ))
if difference < 0.001:
print('Good! The distance matrices are the same')
else:
print('Uh-oh! The distance matrices are different')
print("use compute_distances_no_loop to calculate the dists.....")
# Now implement the fully vectorized version inside compute_distances_no_loops
# and run the code
dists_two = classifier.compute_distances_no_loops(X_test)
# check that the distance matrix agrees with the one we computed before:
print("test the difference between two loop and no loop........")
difference = np.linalg.norm(dists - dists_two, ord='fro')
print('Difference was: %f' % (difference, ))
if difference < 0.001:
print('Good! The distance matrices are the same')
else:
print('Uh-oh! The distance matrices are different')
# Let's compare how fast the implementations are
def time_function(f, *args):
"""
Call a function f with args and return the time (in seconds) that it took to execute.
"""
# you haven't seen it before, the Frobenius norm of two matrices is the square # root of the squared sum of differences of all elements; in other words, reshape # the matrices into vectors and compute the Euclidean distance between them. difference = np.linalg.norm(dists - dists_one, ord='fro') print 'Difference was: %f' % (difference, ) if difference < 0.001: print 'Good! The distance matrices are the same' else: print 'Uh-oh! The distance matrices are different' # In[ ]: # Now implement the fully vectorized version inside compute_distances_no_loops # and run the code dists_two = classifier.compute_distances_no_loops(X_test) # check that the distance matrix agrees with the one we computed before: difference = np.linalg.norm(dists - dists_two, ord='fro') print 'Difference was: %f' % (difference, ) if difference < 0.001: print 'Good! The distance matrices are the same' else: print 'Uh-oh! The distance matrices are different' # In[ ]: # Let's compare how fast the implementations are def time_function(f, *args): """
mask = range(num_test) x_test = x_test[mask] y_test = y_test[mask] # the image data has three chanels # the next two step shape the image size 32*32*3 to 3072*1 x_train = np.reshape(x_train, (x_train.shape[0], -1)) x_test = np.reshape(x_test, (x_test.shape[0], -1)) print 'after subsample and re shape:' print 'x_train : ', x_train.shape, " x_test : ", x_test.shape # KNN classifier classifier = KNearestNeighbor() classifier.train(x_train, y_train) # compute the distance between test_data and train_data dists = classifier.compute_distances_no_loops(x_test) # each row is a single test example and its distances to training example print 'dist shape : ', dists.shape plt.imshow(dists, interpolation='none') plt.show() y_test_pred = classifier.predict_labels(dists, k=5) num_correct = np.sum(y_test_pred == y_test) acc = float(num_correct) / num_test print 'k=5 ,The Accurancy is : ', acc # Cross-Validation # 5-fold cross validation split the training data to 5 pieces num_folds = 5