Why Regularization Isn’t Enough: A Better Way to Train Neural Networks with Two Objectives | Towards Data Science

Why splitting your objectives and your model might be the key to better performance and clearer trade-offs in deep learning.