Propagate ExampleSet

_paul_

Hi,

I'd like to find a good parameter combination using GridParameterOptimization, then write
the parameters to disk (for logging purposes), read them again, set them to a learner and
finally write this optimized model to disk.

Here is the process:

<参数键= " C " value = " operator.Training.parameter.C"/>

The problem is the missing example set for the learner "Final". Can I somehow propagate the
ExampleSet that is also used for the GridParameterOptimization to the second learned ("Final")?

Regards,
Paul

land

Hi Paul,
there are three solutions for this problem. You could either switch on the "keep_example_set" parameter in the XValidation, which will work in this situation. A more generall solution would be to, copy the example set, so that a copy of your exampleset remains, if the first one is consumed somewhere. The following process will show how to do that:

<参数键= " C " value = " operator.Training.parameter.C"/>

A still more sophisticated method would be to use the IOStorage mechanism. With this you can store the data somewhere and retrieve it lateron, even if all IOObjects are thrown away or consumed during the process. Here's how it would work:

<参数键= " C " value = " operator.Training.parameter.C"/>

Greetings,
Sebastian

_paul_

Thank you Sebastian.

This is exactly what I was looking for. :-)

Regards,
Paul