A technique to normalize the weights in each convolutional layer which, when used with batch normalization, can lead to significant performance improvements. 27.07.2023 17:54 aior