In the previous exmaple, we have seen that the split ratio has an impact on the computational time. But the most time-consuming task is optimizing the hyper parameters. Up to now, the
In the previous example, we have seen that the split ratio has an impact on the computational time.
But the most time-consuming task is optimizing the hyper parameters.
Up to now, the platform used the hyper-parameters values given in the config file.
This happens only if one knows the optimal combination of hyper-parameter for the given task.
However, most of the time, they are unknown to the user, and then have to be optimized by the platform.
In this example, we will use the hyper-parameter optimization method implemented in the platform, to do so we will use three lines of the config file :
- ``hps_type:``, controlling the type of hyper-parameter search,
- ``hps_iter:``, controlling the number of random draws during the hyper-parameter search,
- ``nb_folds:``, controlling the number of folds in the cross-validation process.
So if you run ``example 2.2.1`` with :
.. code-block:: python
>>> from multiview_platform.execute import execute
>>> execute("example2.2.1")
The ``hps_type`` argument is set to ``"randomised_search"``, which is at the moment the only hyper-parameter optimization method implemented in the platform.
The ``hps_iter`` argument is set to ``5``,
The ``nb_folds`` argument is set o ``5``.
The computing time should be longer than the previous examples. Let's see the pseudo code of the benchmark, while using the hyper-parameter optimization::
for each monoview classifier:
for each view:
┌
|for each draw (here 5):
| for each fold (here 5):
| learn the classifier on 4 folds and test it on 1
| get the mean performance
|get the best hyper-parameter set
└
learn on the whole training set
and
for each multiview classifier:
┌
|for each draw (here 5):
| for each fold (here 5):
| learn the classifier on 4 folds and test it on 1
