MATLAB program is required.

Problem 6 (18 pts). Load prob6 mat found in Files into your Matlab workspace. It contains a trained neural network, diginset, two images (imgl and img2) and imdsValidation, which is an image data store. Digitnet outputs a digit (from 0 to 9 ) for a given image. Answer the following queitions: a. How many convolution layers there are in digitnat? b. What are digitnet's predictions for imgl and img2? For the remaining part, you need to modify the flle names in imdsValidation to match the fles on your computer. To do this, you need to download the DigitDataset folder used by Matab if you do not have it already on your computer. Locate the folder where your digit images are stored, let's say your images are stored in folder_gart1 DigitDataset... The following script can be used to modify the imdsValidation file names: myloider = 'C:Progran Filesh..nndatasetsl'; \% change to your folder nfile2val = numel(imdsValidation.Fles): for ii = 1:nfile2val; file_qz = imdsValidation.Files[ii); \% fle name on QZs computer beginpos = strfind(file_q2, 'DigitDataset): \% beginning position in string imdsValidation. Fles[i] = [myfolder, fle_qz(beginposend]): if rem(ii, 500)==0, ii, indsValidationfiles[i], end end The above script may take a few minutes to run. If you mess up in changing the files, you can reload prob6.mat. c. What is the overall accuracy of digitnet when it is used to classify the images in imdsValidation (i.e, total number of correct predictions divided by total number tried)? d. Display the confusion chart. Which digit is most accurately predicted? Which digit is least accurately predicted