CS 557: STATISTICAL PATTERN RECOGNITION AND LEARNING
DUE: October 17, 2015.
PROBLEM
1. Read the datasets in the file ‘music.mat’. This is a subset of a dataset of music symbols taken from:
http://www.cvc.uab.es/~afornes/datasets/datasets.htm
Please see Fornés' paper cited at the end for more details:
These images are gray scale images and have all been rescaled to 20x20 images. You can view an image as (e.g., at row 1):
X = trainX(1,:);
im=reshape(X,20,20);
imagesc(im);
You have trainX and trainY for tuning your model.  You have to make submissions on the test set.
3. You have to apply LDA, LLE and MDS to this dataset and use any classifier of your choice to make predictions.
REFERENCES
A.Fornés, J.Lladós, G. Sanchez, "Old Handwritten Musical Symbol Classification by a Dynamic Time Warping Based Method", in Graphics Recognition: Recent Advances and New Opportunities. Liu, W. and Lladós, J. and Ogier, J.M. editors, Lecture Notes in Computer Science, Volume 5046, Pages 51-60, Springer-Verlag Berlin, Heidelberg, 2008.