DisTools examples: Semi-supervised Learning

Labels of objects used for representation, the representation set, are not used in building the dissimilarity space. Consequently, test objects or entirely different objects may be included in this set. Some examples are shown. It is assumed that readers are familiar with PRTools and will consult the following pages where needed:

If the representation set is extended with other objects the dimensionality of the dissimilarity space grows. Some classifiers may profit from this. others might detoriate. Results may depend on the size of the training set. The following experiments may help to study this topic.

Experiment 1

  1. Take a dissimilarity based dataset.
  2. Choose one or more classifiers to be studied.
  3. Compute learning curves by clevald for different choices of the representation set.
  4. Plot all results by plote. Note that several curves can be combined in one plot by plote({e1,e2,e3}), but title and legend should be adapted.

Experiment 2

  1. Take a dissimilarity based dataset.
  2. Choose one or more classifiers to be studied
  3. Compute feature curves by clevalfs using protselfd(‘random’) as a feature selector. Note that the the number of repetitions only affects the choice of the training set. The representation is constant, except for its size.

Are there classifiers / datasets / training set sizes in which one really profits from a larger representation set than the size of the training set?

elements: datasets datafiles cells and doubles mappings classifiers mapping types.
operations: datasets datafiles cells and doubles mappings classifiers stacked parallel sequential dyadic.
user commands: datasets representation classifiers evaluation clustering examples support routines.
introductory examples: Introduction Scatterplots Datasets Datafiles Mappings Classifiers Evaluation Learning curves Feature curves Dimension reduction Combining classifiers Dissimilarities.
advanced examples.

 

Print Friendly, PDF & Email