..  _dimension_reduction:

Dimension Reduction
=====================

The dimension reduction listens on specified scopes and reduces the information to down to a given dimension. Kernel t-SNE to reduce the dimension.

Before the dimension reduction can be used it first has to be trained. The training must be done on a dataset (rsbagfile or livesystem) that has meaningfull entries. So don't train on a set where nothing happens.

Related resources
-----------------

Component repository:
 - Browse compenent repository `dimension-reduction <https://projects.cit-ec.uni-bielefeld.de/projects/lsp-csra/repository/dimension-reduction>`_.

Clone repositories:
 - e.g. ``git clone https://projects.cit-ec.uni-bielefeld.de/git/lsp-csra.dimenson-reduction.git``

System startup:
 - ``dimensionreduction`` 

System help:
 - ``dimensionreduction --help``

Related projects:
 - t-SNE demonstration http://distill.pub/2016/misread-tsne/

Interfaces
-----------------


=========================================== ========================================
Scope (Informer)                            Type
=========================================== ========================================
 ``/dimensionreduction``                     `Point2D <https://projects.cit-ec.uni-bielefeld.de/projects/rst-experimental/repository/revisions/master/entry/proto/experimental/rst/dimensionreduction/Point2D.proto>`_
=========================================== ========================================


Examples
--------------

Training the Dimension Reduction component on a given rsbag file:

.. code-block:: bash
   :emphasize-lines: 4,6,7

    dimension_reduction PARAMETERS
    dimenison_reduction -t
    dimenison_reduction -t -k <kernelFile>
    dimenison_reduction -t -k kernel.ser
    dimenison_reduction -t -r <rsbagFile>
    dimenison_reduction -t -r recording.tide
    dimenison_reduction -t --gui

Run Dimenison Reduction componont after training

.. code-block:: bash
   :emphasize-lines: 4,6,7

    dimension_reduction PARAMETERS
    dimenison_reduction
    dimenison_reduction -k <kernelFile>
    dimenison_reduction -k kernel.ser
    dimenison_reduction -r <rsbagFile>
    dimenison_reduction -r recording.tide
    dimenison_reduction --gui