Achieving faithful explainability in feedforward neural networks through accurately computed feature attribution | Publicación