fedbox.models
This submodule defines the two main statistical models to be used in the simulations.
- MultinomialLogisticRegression is a multinomial logistic regression model that allows testing in a smooth and strongly convex setting.
- NeuralNetwork is a simple ReLU-activated neural network with one hidden layer. This allows testing in a nonconvex scenario.
1''' 2This submodule defines the two main statistical models to be used in the simulations. 3<ol> 4 <li><b>MultinomialLogisticRegression</b> is a multinomial logistic regression model that allows testing in a 5 smooth and strongly convex setting.</li> 6 <li><b>NeuralNetwork</b> is a simple ReLU-activated neural network with one hidden layer. This allows testing in 7 a nonconvex scenario.</li> 8</ol> 9'''