Implement in Python the Naive Bayes model discussed in lecture for the MNIST dataset. h(x) = argmax_kappa epsilon{0, 1, ..., 9} [fx| kappa(x)P[Y = k]]. = argmax_kappa elementof{0, 1, ..., 9}[log(f_x|kappa(x)) + log(P[Y = k])] (1) The function h(x) is the model prediction for the image x. f_X|kappa(x) is the conditional joint probability density function of X given Y = k. Let X/Y = k ~ N(mu^kappa, sigma^kappa). That is conditional on Y = k, X is a multi-dimensional Gaussian random variable with mean mu^kappa elementof R^784 and convariance matrix sigma^kappa elementofR^784 times784.Estimate P[Y = k], mu^k, and sigma^kappa from the training set. A "regularizer" should be added. to sigma^kappa. That is, replace sigma^kappa with sigma^~^kappa = sigma^kappa + lambda = 1/10. Calculate the accuracy of the model prediction h(x) using the regularized covariance matrix sigma^~^k on the test set. Write a Python function to display the first K images from the MNIST test set, the model predictions, and the correct labels. Implement in Python the Naive Bayes model discussed in lecture for the MNIST dataset. h(x) = argmax_kappa epsilon{0, 1, ..., 9} [fx| kappa(x)P[Y = k]]. = argmax_kappa elementof{0, 1, ..., 9}[log(f_x|kappa(x)) + log(P[Y = k])] (1) The function h(x) is the model prediction for the image x. f_X|kappa(x) is the conditional joint probability density function of X given Y = k. Let X/Y = k ~ N(mu^kappa, sigma^kappa). That is conditional on Y = k, X is a multi-dimensional Gaussian random variable with mean mu^kappa elementof R^784 and convariance matrix sigma^kappa elementofR^784 times784.Estimate P[Y = k], mu^k, and sigma^kappa from the training set. A "regularizer" should be added. to sigma^kappa. That is, replace sigma^kappa with sigma^~^kappa = sigma^kappa + lambda = 1/10. Calculate the accuracy of the model prediction h(x) using the regularized covariance matrix sigma^~^k on the test set. Write a Python function to display the first K images from the MNIST test set, the model predictions, and the correct labels