What’s in the Box?
      Exploring the Inner Life of Neural Networks with Robust Rules

                   Jonas Fischer 1 Anna Olah 1 Jilles Vreeken 2

            Abstract                 Y are typically active when neurons X are. For robustness
                                 we explicitly allow for noise, and to ensure that we dis-
   We propose a novel method for exploring how
                                 cover a succinct yet descriptive set of rules that captures the
   neurons withi ...