Federated Sklearn

Much like making a Federated PyTorch, making custom SGD models gives the user the ability to customize and extend OctaiPipe’s model implementation to tailor more specifically to each user’s problem. This allows users to develop a custom solution and integrate it seamlessly in the OctaiPipe framework.

This guide will not go into great detail of custom models as this is covered in Federated PyTorch. Instead, the base classes for the SGD classifier and regressor are shown below, and the same structure for extending these classes as explained for PyTorch models can be applied to the SGD models.

Base Classes

class octaipipe_lite.model_classes.fl_aquarium.base_SGDClassifier.BaseSGDClassifier(input_shape: int = None, output_shape: int = None, metric: str = None, **kwargs)

class octaipipe_lite.model_classes.fl_aquarium.base_SGDRegressor.BaseSGDRegressor(input_shape: int = None, output_shape: int = None, metric: str = None, **kwargs)