Pearl: Prototype lEArning via Rule Lists

Tianfan Fu*, Tian Gao*, Cao Xiao*, Tengfei Ma*, Jimeng Sun

Abstract: Deep neural networks have demonstrated promising prediction and classification performance on many healthcare applications. However, the interpretability of those models are often lacking. On the other hand, classical interpretable models such as rule lists or decision trees do not lead to the same level of accuracy as deep neural networks and can often be too complex to interpret (due to the potentially large depth of rule lists). In this work, we present PEARL, Prototype lEArning via Rule Lists, which iteratively uses rule lists to guide a neural network to learn representative data prototypes. The resulting prototype neural network provides accurate prediction, and the prediction can be easily explained by prototype and its guiding rule lists. Thanks to the prediction power of neural networks, the rule lists from prototypes are more concise and hence provide better interpretability. On two real-world electronic healthcare records (EHR) datasets, PEARL consistently outperforms all baselines across both datasets, especially achieving performance improvement over conventional rule learning by up to 28% and over prototype learning by up to 3%. Experimental results also show the resulting interpretation of PEARL is simpler than the standard rule learning.

Keywords: rule list learning, prototype learning, interpretability, healthcare

TL;DR: a method combining rule list learning and prototype learning

Data: [MIMIC-III](https://paperswithcode.com/dataset/mimic-iii)