Batch and Layer Normalization

Recent advances in deep learning research have revolutionized fields like medical imaging, machine vision, and natural language processing. However, it’s still challenging for data scientists to choose the optimal model architecture and to tune hyperparameters for best results.


This is a companion discussion topic for the original entry at https://www.pinecone.io/learn/batch-layer-normalization/

Hi, just wanted to note that in general Standardization does not transform input values such that they follow a normal distribution, as it is just a linear transformation.

Hi, thank you for pointing out. I realize that I haven’t made assumptions on the input distribution to state it that way. The post has been updated for clarity.