All translations

Enter a message name below to show all available translations.

Found 3 translations.

Name	Current message text
^h English (en)	Deep neural networks with many parameters are powerful function approximators but are prone to {{Term\|overfitting}}, especially when training data is limited. Traditional {{Term\|regularization}} methods such as L2 {{Term\|weight decay}} and early stopping provided some relief, but were often insufficient for large networks. Model combination — training multiple models and averaging their predictions — was known to reduce {{Term\|overfitting}} but was computationally expensive.
^h Spanish (es)	Las redes neuronales profundas con muchos parámetros son aproximadores de funciones potentes, pero son propensas al {{Term\|overfitting\|sobreajuste}}, especialmente cuando los datos de entrenamiento son limitados. Los métodos tradicionales de {{Term\|regularization\|regularización}}, como el {{Term\|weight decay\|decaimiento de pesos}} L2 y la parada temprana, proporcionaban cierto alivio, pero a menudo resultaban insuficientes para redes grandes. La combinación de modelos —entrenar múltiples modelos y promediar sus predicciones— era conocida por reducir el {{Term\|overfitting\|sobreajuste}}, pero resultaba computacionalmente costosa.
^h Chinese (zh)	具有大量参数的深度神经网络是强大的函数逼近器，但容易出现{{Term\|overfitting\|过拟合}}，尤其是在训练数据有限时。传统的{{Term\|regularization\|正则化}}方法，例如 L2 {{Term\|weight decay\|权重衰减}}和提前停止，可以提供一定缓解，但对于大型网络往往不够充分。模型组合——训练多个模型并对它们的预测取平均——已知可以减少{{Term\|overfitting\|过拟合}}，但计算成本高昂。