This distribution contains three dimensionality reduction programs. 1. The first is 2pwmv that implements the supervised dimensionality reduction algorithm described in "Weighted maximum variance dimensionality reduction" by Turki Turki and Usman Roshan. Usage: ./2pwmv where = number of rows in data file = number of columns in data file = datafile format described below = desired number of eigenvectors = alpha parameter of algorithm (see paper), recommended set to 1 = beta parameter of algorithm (see paper), recommended to cross-validate from values (-2, -1.9, -1.8, -1.7, -1.6, -1.5, -1.4, -1.3, -1.2, -1.1, -1, -.9, -.8, -.7, -.6, -.5, -.4. -.3, -.2, -.1, -.01) = filename = file containing labels (format described below) The standard output are the k leading eigenvectors which can be redirected to a file. 2. The second program is the weighted maximum margin criterion (also supervised) described in "Weighted maximum margin discriminant analysis with kernels" by Zheng, W., Zou, C. and Zhao, L. in Neurocomputing 67 (2005) pages 357-362. Usage: ./wmmc where = number of rows in data file = number of columns in data file = datafile format described below = desired number of eigenvectors = weight parameter of algorithm, recommended to cross-validate from values (10^-7, 10^-6, 10^-5, 10^-4, 10^-3, .01, .1, 1, 10, 100) = filename = file containing labels (format described below) 3. The third is the popular principal component analysis algorithm (unsupervised). Usage: ./pca where = number of rows in data file = number of columns in data file = datafile format described below = desired number of eigenvectors = filename = file containing labels (format described below) ------------------------------------------------------------------------------ Data file format 1. Each row is a vector of entries separated by space. All entries including 0 must be specified. This is not a sparse format. 2. Each row contains two numbers. The first number is the class that is denoted by a natural number. The second is a row number i that refers to the vector in row i in the datafile. As an example below is a dataset of four vectors of dimensionality three where the first two belong to class 0 and remaining to class 1. 10 2 4 12 1 5 2 1 0 1 2 1 The example label file would be 0 0 0 1 1 2 1 3