Problem 1: Digit classification (30 Marks) In this problem you will classify digits from small handwritten images. 1. (5

[answerhappygod]

Problem 1 Digit Classification 30 Marks In This Problem You Will Classify Digits From Small Handwritten Images 1 5 1 (386.19 KiB) Viewed 29 times

Problem 1: Digit classification (30 Marks) In this problem you will classify digits from small handwritten images. 1. (5 marks) Use principle components analysis to produce a 5 dimensional feature vector for each 64 dimensional digit image. 2. (5 mark) Split your low dimensional data into training and test sets. 3. (10 marks) Fit a logistic regression classifier to the training set and estimate the the predictive power of the model using the test set. Plot a bar chart showing the prediction accuracy for each digit. 4. (10 mark) Open ended question: Using any method you wish, build a digit classifier with the best possible predictive power. Credit will be given for for clear coding and comments, creative and rigourous use of methods, and quality of predictions on the test data. The cell below will load the data in the form a set of image vectors X and digit values y. = #First download the image data and plot some examples digits = load_digits() X = digits.data y = - digits.target #The data consists of 8x8 pixel images unravelled into vectors of length 64 #To plot a digit you must reshape into and 8x8 array #Example fig, ax = plt. subplots(1,5, figsize=( 10,2)) for i in range(5): dig = X. reshape(8,8) ax.imshow(dig, cmap='Greys') ax.axis('off') plt.show() =

中1234