4.8. Optimizers¶

Optimizers change the parameters such that the value of the loss function is minimized. ParAMS provides a base class wrapper for different optimization algorithms, making the selection of this component open-ended. An optimizer instance can be passed to the Optimization class, once everything else is set up:

>>> o = CMAOptimizer()
>>> optimization = Optimization(jobcol, training_set, interface, o)
>>> optimization.optimize()

Currently, the following optimization algorithms are included: