Коэффициент регуляризации — гиперпараметр, управляющий силой регуляризации во время обучения. Его увеличение может снизить overfitting, но также может уменьшить predictive power, если модель становится слишком ограниченной.
A number that specifies the relative importance of regularization during training. Raising the regularization rate reduces overfitting but may reduce the model's predictive power. Conversely, reducing or omitting the regularization rate increases overfitting. The regularization rate is usually represented as the Greek letter lambda. The following simplified loss equation shows lambda's influence: where regularization is any regularization mechanism, including; - L~1~ regularization - L~2~ regularization See Overfitting: L2 regularization in Machine Learning Crash Course for more information.