A Neural Network That Can Explain It's Inferences

Here's an article that explains the methodology and model of the paper. The system was developed by MIT researchers and is called "Transparency by Design network (TbD-Net)":

https://towardsdatascience.com/transparent-reasoning-how-mit...