For an image, we use a superpixel segmenter, which selects regions on the image. For text, we use the natural correspondence between an input embedding and a word token. Similar notions of input encodings have also been used in [39, 48]. Based on the definition of non-additive statistical interaction (Def. Based on Eqs. 7 - 9 of Lemma 2: φ (I SOC does not assign attributions to general feature sets, only contiguous feature sequences.

Many functions characterising physical systems are additively separable. This is the case, for instance, of mechanical Hamiltonian functions in physics, population growth equations in biology, and consumer preference and utility functions in economics. We consider the scenario in which a surrogate of a function is to be tested for additive separability. The detection that the surrogate is additively separable can be leveraged to improve further learning. Hence, it is beneficial to have the ability to test for such separability in surrogates. The mathematical approach is to test if the mixed partial derivative of the surrogate is zero; or empirically, lower than a threshold. We present and comparatively and empirically evaluate the eight methods to compute the mixed partial derivative of a surrogate function.