6688

Fashion-MNISTis a dataset of Zalando’s article images — consisting of a training set of 60,000 examples and a test set of 10,000 examples. Each example is a 28×28 grayscale image, associated with a label from 10 classes. Fashion- MNIST is a replacement for the original MNIST dataset (handwritten digits) for benchmarking machine learning algorithms. It shares the same image size and structure of training and testing splits. The web pagehttps://github.com/ zalandoresearch/fashion-mnisthas more information.

Document Preview:

Predictive Analytics Assignment — 2018SubmissionThe assignment solution should be submitted electronically, and can be a combinationof R code and PDF document, by email to O.Obst@westernsydney.edu.au.Include the completed cover sheet that you can ?nd at the end of the document.Submission is due on 18 Nov 2018, 11:59pm.1. Fashion-MNIST is a dataset of Zalando’s article images — consisting of atraining set of 60,000 examples and a test set of 10,000 examples. Each exampleis a 28×28 grayscale image, associated with a label from 10 classes. Fashion-MNIST is a replacement for the original MNIST dataset (handwritten digits) forbenchmarking machine learning algorithms. It shares the same image size andstructure of training and testing splits. The web pagehttps://github.com/zalandoresearch/fashion-mnist has more information.Figure 1: An example for how some the data looks like (each class takes three-rows).1In the vuws data directory for the assignment is a zip archive (fashionmnist.zip)that contains 4 data ?les (training and test sets, split into data and labels), plussome R code (loader.R).You can use the functions inloader.R to load the unzipped ?les, and also to dis-1play individual images while you work on your code . The functionload mnist()in this ?le loads all the data (must be in the same directory), and creates variablestrain$n,train$x,train$y, andtest$n,test$x,test$y.After loading,train$x is a 60000 x 784 matrix, each row is one digit (28×28).thYou can use the callshow digit(train$x[5,]) to display the5 trainingexample. The labels (0–9) for the data are in the train$y and test$y vari-ables.(a) Using the nn2 function in the RANN library, implement a 3 nearest neigh-bour classi?cation. Submit your R code, and the achieved accuracy onthe test set (accuracy: proportion of correct results from total number ofclassi?cations).(b) Thenn2 function also returns the distance to thek nearest neighbours (fordetails, see thenn2 help text)….

Attachments:

Assignment.pdf