Free Trial

Safari Books Online is a digital library providing on-demand subscription access to thousands of learning resources.

298 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C HAPTER 14 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 0 0 0 0 0 0 1 1 1 1 0 0 0 0 0 0 Simplifying data with the singular value decomposition 1 1 1 1 1 1 1 0 0 0 1 1 1 1 1 1 1 0 0 0 0 0 0 0 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 0 0 0 0 0 1 1 1 1 0 0 0 0 0 0 1 1 1 0 0 0 1 1 1 1 1 0 0 0 0 0 1 1 1 1 0 0 0 0 0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 With as few as two singular values, the image is reconstructed quite accurately. How many numbers did we use to reconstruct this image? Each of the U and V T matrices was 32x2, and there were two singular values. That's a total 64+64+2 = 130. Compare this with the original number, which was 1024, and you get an almost 10x compression. 14.7 Summary The singular value decomposition ( SVD ) is a powerful tool for dimensionality reduc- tion. You can use the SVD to approximate a matrix and get out the important features. By taking only the top 80% or 90% of the energy in the matrix, you get the important features and throw out the noise. The SVD is employed in a number of applications today. One successful application is in recommendation engines.