Classification by computer of handwritten digits is a standard problem in pattern recognition.

computer science

Description

Problem Statement 

Classification by computer of handwritten digits is a standard problem in pattern recognition. The typical application is automatic reading of zip codes on envelopes. In this assignment you’ll address the following problem: Given a set of manually classified digits (the training set), classify a set of unknown digits (the test set) using SVD method.


Data Set 

We will be using the US postal Service database that contains 1707 training and 2007 test digits (uploaded to Canvas). Each image is a grayscale 16x16 image that is converted to a 256x1 column vector by stacking all the columns of each image matrix above each other.


• The training images are stored in trainInput.csv. (256x1707).

 • The correct digit corresponding to each column of trainInput is stored in trainOutput.csv. (1x1707). 

• The test images are stored in testInput.csv. (256x2007).

 • The correct digit corresponding to each column of testInput is stored in testOutput.csv. (1x2007).


Related Questions in computer science category