Finding Interpretable Class-Specific Patterns through Efficient Neural Search

Nils Philipp Walter, Jonas Fischer, Jilles Vreeken

Abstract:

Discovering patterns in data that best describe the differences
between classes allows to hypothesize and reason about class-
specific mechanisms. In molecular biology, for example, this
bears promise of advancing the understanding of cellular pro-
cesses differing between tissues or diseases, which could lead
to novel treatments. To be useful in practice, methods that
tackle the problem of finding such differential patterns have
to be readily interpretable by domain experts, and scalable to
the extremely high-dimensional data.
In this work, we propose a novel, inherently interpretable bi-
nary neural network architecture DIFFNAPS that extracts dif-
ferential patterns from data. DIFFNAPS is scalable to hun-
dreds of thousands of features and robust to noise, thus over-
coming the limitations of current state-of-the-art methods in
large-scale applications such as in biology. We show on syn-
thetic and real world data, including three biological appli-
cations, that, unlike its competitors, DIFFNAPS consistently
yields accurate, succinct, and interpretable class descriptions.