Sampling: creation of emails used to inspect the model Scope x e = "Hello, I am very happy to be at SRA. Clément Henin, PhD student at Inria" Samples s 1 = "Hello, I am very happy to be at SRA. Clément Henin, PhD student at Inria" s 2 = "Hello, I am very happy to be at SRA. Alan Turing" s 3 = "Hello, I am very happy to be at SRA. Claude Shannon" s 4 = "Hello, I am very happy to be at SRA. frank rosenblatt" SAMPLING Population x 1 = "Hello, If a machine is expected to be infallible, it cannot also be intelligent. Alan Turing" x 2 = "Hello, Information is the resolution of uncertainty. Claude Shannon" x 3 = "Hello, The theory has been developed for a hypothetical nervous system, or machine, called a perceptron. frank rosenblatt" Clément Henin & Daniel Le Métayer A Multi-layered Approach for Interactive Black-box Explanations