Q$150$ points $5$ pts each for $1\mathrm{.}1-1\mathrm{.}5$ and $25$ pts for Q$1\mathrm{.}61.$ Load MNIST dataset$2.$ Take one random image from each class and estimate L$1$ distance to every other class. Display theseresults as cell output$3.$ Load CIFAR$10$ dataset$4.$ Take one random image from each class and estimate L$1$ distance to every other class. Display theseresults as cell output$5.$ Find the closest neighboring class $($K$=1)$ based on L$1$ distance for each image class and provide yourthoughts on why these images from different class looks similar$\mathrm{.}6.$ Repeat Steps $1-5$ but replace L$1$ distance with L$2$ distanceQ$250$ points $5$ pts each for $2\mathrm{.}1-2\mathrm{.}4$ and $30$ pts for Q$2\mathrm{.}51.$ Divide the MNIST data in train $(80\%)$ and test $(20\%)2.$ Based on L$1$ distance function, run K$-$nearest neighbor classifier for K$=5$ on test set$\mathrm{.}3.$ Estimate the computing time to complete testing$4.$ Repeat steps $2$ and $3$ for K$=25,55,$ and $1055.$ Repeat steps $2,3,$ and $4$ with L$2$ as a distance functionQ$350$ points $10$ pts each$1.$ Implement the same code in the link provided in Resources section $($R$3)$ for MNIST dataset$2.$ Repeat step $1$ by changing the number of neurons in the second layer to same size as your input$3.$ Repeat step $1$ by changing the number of neurons in the second layer to $2$ times the size of your input$4.$ Repeat step $1$ by adding another dense layer with $($your chosen number$)$ neurons$5.$ Compare the accuracy for the $3$ above classifiers from $1-4$Q$420$ points$1.$ Summarize the key observations in Q$1,$ Q$2,$ and Q$3.$Bonus Q $30$ points $15$ pts each$1.$ Apply KNN to fashion MNIST data $($provided in the link R$1\mathrm{.}3)2.$ Compare accuracies of MLP on fashion MNIST and KNN from Bonus Q$1.$Resources$1.$ To load MNIST and CIFAR$10$ dataset from Tensorflow and KerasModule: tf$.$keras.datasets $|$ TensorFlow v$2\mathrm{.}15\mathrm{.}0.$post$12.$ Implementing KNN from scratchImplementing KNN from Scratch $|$ Nikita Kozodoi$3.$ Multi$-$layer Perceptron for image classificationBasic classification: Classify images of clothing $|$ TensorFlow Core