SPIDER The Spider Objects

Multi-class feature selection using spectral clustering

   A=CLUSTUB(H) returns a clustub object initialized with hyperparameters H. 
    Peforms feature selection via spectral clustering.
   Hyperparameters, and their defaults
    feat=[]              -- number of features
    output_rank=0        -- set to 1 if only the feature ranking matters
                            (does not perform any classification on the data)
    child=spectral('k=2')-- Clustering method used for training
    w                    -- the weights
    b0                   -- the threshold (when using all features)
    rank                 -- the ranking of the features
    d                    -- training set
    train, test, get_w 
    d=gen(bayes({gauss([-1 3]) gauss([0 4]) gauss([1 2])})) 
    Method for multi-class feature selection.
    To use with SVM, use:  

the Emergence of Sparsity in a Weighted-based Approach.
Reference : Feature Selection for Unsupervised and Supervised Inference:
Author : Lior Wolf and A. Shashua
Link : http://www.cs.huji.ac.il/~shashua/papers/fts-long.pdf