Integrating OneDAL primitives for `LogisticRegression`

[avolkov-intel]

Description

Add scikit-learn-intelex interface for oneDAL algorithm objective_function

• Defined new class LogisticLoss (defined in onedal/objective_function/objective_function.py). Its interface is almost alligned to LinearModelLoss class from stock scikit-learn https://github.com/scikit-learn/scikit-learn/blob/7db5b6a98ac6ad0976a3364966e214926ca8098a/sklearn/linear_model/_linear_loss.py#L9
• Defined functions loss, gradient, loss_gradient, gradient_hessian, gradient_hessian_product
• Covered new functionality with tests

[samir-nasibli]

I really don't understand which api are your implementing here? in vanilla scikit-learn we already have API for log_loss https://scikit-learn.org/stable/modules/generated/sklearn.metrics.log_loss.html . I think it's worth following this API

[avolkov-intel]

I really don't understand which api are your implementing here? in vanilla scikit-learn we already have API for log_loss https://scikit-learn.org/stable/modules/generated/sklearn.metrics.log_loss.html . I think it's worth following this API

I implement the API for LinearModelLoss class. https://github.com/scikit-learn/scikit-learn/blob/7db5b6a98ac6ad0976a3364966e214926ca8098a/sklearn/linear_model/_linear_loss.py#L9 This class is used internally by scikit-learn LogisticRegression

[avolkov-intel]

@KulikovNikita @Alexsandruss could you review the code, please?

[Alexsandruss]

/intelci: run

[samir-nasibli]

/intelci: run

[samir-nasibli]

Please rebase and run intel CI if it is required

[avolkov-intel]

/intelci: run

[ethanglaser]

Can reopen if progress continues