The Only Way to make Deep Learning Interpretable is to Have it Explain Itself

One of the great biases that Machine Learning practitioners and Statisticians have is that our models and explanations of the world should be parsimonious. We’ve all bought into Occam’s Razor…

Actions
Connections