z=WTx X, y = arg max zi - 1. 1≤i≤10 Calculate the accuracy, show the confusion matrix.
We have mnist 10 testing and 10 training file.
Train the unthresholded perceptron for AND function with the least squares error using
a) stochastic gradient descent
b) nonstochastic gradient descent
Use the learning rate η = 0.2. Start with initial w = (w0,w1,w2) = (0, 0, 0).
Display w after each update.
• Train the unthresholded 10 perceptrons to recognize MNIST handwrit- ten digits. Label digits from each class with corresponding numeric values:
(1,0,0,0,0,0,0,0,0,0) for 0, (0,1,0,0,0,0,0,0,0,0) for 1, (0,0,1,0,0,0,0,0,0,0) for 2, ...
(0,0,0,0,0,0,0,0,0,1) for 9.
Then find W using
a) stochastic gradient descent
b) nonstochastic gradient descent
Take 100 digits from each training file total number of 1000 training examples. Use the least squares error. Use the learning rate η = 0.2. Take initial W = 0. After training classify testing digits using
z=WTx, y=argmaxzi−1. 1≤i≤10
Calculate the accuracy, show the confusion matrix.
Step by step
Solved in 2 steps with 1 images
