iRace vs mlr

[Amirhossein Moosavi]

Hi Everyone,

I would like to know what is the difference between iRace and mlr packages? I can see in the iRace tutorial pdf file that authors recommend mlr for machine learning algorithms. I cannot understand the reason. Thanks in advance for the clarification.

Best, Amirhossein.

[Manuel López-Ibáñez]

Dear Amirhossein,

irace is a tool for configuring (tuning) hyper-parameters of algorithms, optimization algorithms, machine learning models, compilers, database managers, etc.

mlr is a framework for machine learning in R.

You can use irace to tune the machine learning models in mlr: https://mlr3tuning.mlr-org.com/dev/reference/mlr_tuners_irace.html

You can use irace to tune other machine learning packages/models not in mlr (for example, scikit-learn).

You can also use other algorithms to tune the machine learning models in mlr.

I hope the above clarifies the difference between irace and mlr.

Best,

Manuel.