La Normalisation L1, également connue sous le nom de Régularisation L1 or Lasso normalization, is a method used in various fields of apprentissage automatique and analyse de données to scale data and améliorer la performance du modèle. The primary goal of L1 Normalization is to adjust the values of the data points such that their total absolute value equals one. This is achieved by dividing each individual value by the sum of the absolute values of all data points in the dataset.
La formule de la normalisation L1 peut être exprimée comme suit :
Ici, x represents an individual data point, and the denominator is the sum of the absolute values of all data points in the dataset. This process ensures that the transformed data will be bounded within the range of -1 to 1, leading to a more uniform distribution of values.
L1 Normalization is particularly useful in scenarios where the data may have varying scales or units, as it helps to eliminate biases that might arise from such differences. It is commonly used in algorithms such as Lasso regression, where it encourages sparsity in the model by shrinking some coefficients to zero. This characteristic makes L1 Normalization a valuable technique for sélection de caractéristiques dans des ensembles de données à haute dimension.
Dans l'ensemble, la Normalisation L1 joue un rôle crucial dans la préparation de pour des modèles d’apprentissage automatique, ensuring that each feature contributes equally to the final outcome.