Explaining Deep Tractable Probabilistic Models: The sum-product network case
This work addresses the interpretability challenge for domain experts using tractable deep probabilistic models, though it is incremental as it builds on existing SPN frameworks.
The paper tackles the problem of explaining Sum-Product Networks (SPNs) by developing an algorithm called ExSPN that converts SPNs into interpretable context-specific independence trees (CSI-trees), demonstrating superior explainability in evaluations on synthetic, standard, and real-world clinical datasets.
We consider the problem of explaining a class of tractable deep probabilistic models, the Sum-Product Networks (SPNs) and present an algorithm ExSPN to generate explanations. To this effect, we define the notion of a context-specific independence tree(CSI-tree) and present an iterative algorithm that converts an SPN to a CSI-tree. The resulting CSI-tree is both interpretable and explainable to the domain expert. We achieve this by extracting the conditional independencies encoded by the SPN and approximating the local context specified by the structure of the SPN. Our extensive empirical evaluations on synthetic, standard, and real-world clinical data sets demonstrate that the CSI-tree exhibits superior explainability.