Task 1: EigenFaces “Eigenfaces” is the general name given to a dataset of face images after applying PCA to the images. Eigenfaces are the Principal Components for the dataset visualized as images. Remember that principal components these are the new axes of the transformed dataset in the directions of maximum variance. Eigenfaces were introduced in a seminal paper by Turk and Pentland in 1991 and have since been used widely as input for classifiers and face recognition systems. The advantage of eigenfaces is lowering the dimensionality of images, saving computational power and achieving better classification performance. The combination of SVMs and PCA can be superior to neural networks when the number of images available is not very large.
Task 2: Classification of cancer gene expressions In this task, you are asked to develop an SVM classifier for 14 cancer types according to their gene expressions in tissue samples. This dataset was collected and analyzed using several techniques including SVMs in this paper http://www.pnas.org/content/98/26/15149.full. Please read the Abstract of the paper and the section on SVMs. The results section is an optional reading.