Normalisation de groupe
La Normalisation de Groupe (GN) est une technique utilisée dans le domaine de apprentissage profond to normalize the inputs of a réseau neuronal. Unlike traditional normalization methods like Normalisation de lot, which normalizes across the entire batch of data, Group Normalization divides the features into smaller groups and normalizes each group independently.
Cette méthode est particulièrement bénéfique dans les scénarios où la taille du lot is small, such as in tasks involving segmentation or when working with images of varying sizes. By normalizing groups of features rather than the entire batch, Group Normalization can maintain the statistical properties of the activations more effectively.
In practice, Group Normalization works by calculating the mean and variance for each group of features, which are then used to scale and shift the inputs. This process helps in stabilizing the learning process by reducing décalage de covariables interne, allowing for faster convergence during training.
Group Normalization has become increasingly popular in various applications, including computer vision and traitement du langage naturel, as it can enhance model performance and robustness. It provides an alternative to Batch Normalization, especially in settings where batch sizes cannot be large enough to effectively estimate the mean and variance.
In summary, Group Normalization is a valuable tool in deep learning, offering a solution to normalization that adapts well to different batch sizes and helps improve the stability and performance of réseaux neuronaux.