from matplotlib import cm import matplotlib.pyplot as plt import platform import numpy as np from mlxtend.data import loadlocal_mnist from sklearn.datasets import load_digits from sklearn.model_selection import train_test_split import pickle plt.rcParams.update({'font.size': 12}) plt.rcParams['figure.figsize'] = [8, 4] X, y = loadlocal_mnist(images_path='train-images-idx3-ubyte', labels_path='train-labels-idx1-ubyte') # Keeping only digits 0 and 1 index = y <2 X = X[index] y = y[index] # Normalizing the data X = X/X.max() # Splitting to training and testing groups. X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, shuffle=False) def sigmoid(t): reasonable = (t>-100).astype(float) t = t*reasonable s = 1/(1+np.exp(-t)) s = s*reasonable return s Nsamples,Nfeatures = X_train.shape Nclasses = 2 a = np.random.randn(Nfeatures+1,Nclasses-1) Xtilde = np.concatenate((X_train,np.ones((Nsamples,1))),axis=1) gamma = 1e-1 for iter in range(1500): # YOUR CODE HERE a = a - gamma*gradient if(np.mod(iter,100)==0): print("Error = ",np.sum(error**2)/) fig,ax = plt.subplots(1,2) ax[0].plot(s[:,0:200].T) ax[0].plot(y_train[0:200]) ax[0].set_title('True and predicted labels') ax[1].plot(error) ax[1].set_title('Prediction Errors') plt.show() plt.imshow(np.reshape(a[:-1],(28,28))) plt.title("weights") Nsamples,Nfeatures = X_test.shape Xtilde = np.concatenate((X_test,np.ones((Nsamples,1))),axis=1).T s_test = sigmoid(a.T@Xtilde) error = (s_test - y_test).T print(np.sum(error**2)) plt.figure() plt.plot(s_test[:,400:700].T) plt.plot(y_test[400:700]) plt.title("True and predicted labels") plt.show() plt.figure() plt.plot(error) plt.title("Error") wrong_indices = np.where(np.abs(error[:,0])>0)[0] Xwrong = X_test[wrong_indices] fig, axs = plt.subplots(1, len(wrong_indices)) for i in range(len(wrong_indices)): axs[i].imshow(np.reshape(Xwrong[i],(28,28))) axs[i].set_title("Misclassified") This code has only one part missing where your code here is written. I need help with completion of that part. Can someone please help me with the same?
from matplotlib import cm
import matplotlib.pyplot as plt
import platform
import numpy as np
from mlxtend.data import loadlocal_mnist
from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
import pickle
plt.rcParams.update({'font.size': 12})
plt.rcParams['figure.figsize'] = [8, 4]
X, y = loadlocal_mnist(images_path='train-images-idx3-ubyte', labels_path='train-labels-idx1-ubyte')
# Keeping only digits 0 and 1
index = y <2
X = X[index]
y = y[index]
# Normalizing the data
X = X/X.max()
# Splitting to training and testing groups.
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, shuffle=False)
def sigmoid(t):
reasonable = (t>-100).astype(float)
t = t*reasonable
s = 1/(1+np.exp(-t))
s = s*reasonable
return s
Nsamples,Nfeatures = X_train.shape
Nclasses = 2
a = np.random.randn(Nfeatures+1,Nclasses-1)
Xtilde = np.concatenate((X_train,np.ones((Nsamples,1))),axis=1)
gamma = 1e-1
for iter in range(1500):
# YOUR CODE HERE
a = a - gamma*gradient
if(np.mod(iter,100)==0):
print("Error = ",np.sum(error**2)/)
fig,ax = plt.subplots(1,2)
ax[0].plot(s[:,0:200].T)
ax[0].plot(y_train[0:200])
ax[0].set_title('True and predicted labels')
ax[1].plot(error)
ax[1].set_title('Prediction Errors')
plt.show()
plt.imshow(np.reshape(a[:-1],(28,28)))
plt.title("weights")
Nsamples,Nfeatures = X_test.shape
Xtilde = np.concatenate((X_test,np.ones((Nsamples,1))),axis=1).T
s_test = sigmoid(a.T@Xtilde)
error = (s_test - y_test).T
print(np.sum(error**2))
plt.figure()
plt.plot(s_test[:,400:700].T)
plt.plot(y_test[400:700])
plt.title("True and predicted labels")
plt.show()
plt.figure()
plt.plot(error)
plt.title("Error")
wrong_indices = np.where(np.abs(error[:,0])>0)[0]
Xwrong = X_test[wrong_indices]
fig, axs = plt.subplots(1, len(wrong_indices))
for i in range(len(wrong_indices)):
axs[i].imshow(np.reshape(Xwrong[i],(28,28)))
axs[i].set_title("Misclassified")
This code has only one part missing where your code here is written. I need help with completion of that part. Can someone please help me with the same?
It seems that the missing part involves computing the gradient of the binary logistic regression loss function with respect to the weight parameter vector a.
Trending now
This is a popular solution!
Step by step
Solved in 5 steps with 3 images
