RAPID: Rapid and Precise Interpretable Decision Sets

Interpretable Decision Sets (IDS) is an approach to building transparent and interpretable supervised machine learning models. Unfortunately, IDS does not scale to most commonly encountered big data sets. In this paper, we propose Rapid And Precise Interpretable Decision Sets (RAPID), a faster alternative to IDS. We use the existing formulation of decision set learning and propose a time-efficient learning framework. RAPID has two major improvements over IDS. First, it uses a linear-time randomized Unconstrained Submodular Maximization algorithm to optimize the objective function. Second, we design special data structures, based on Frequent-Pattern (FP) trees to achieve better computational efficiency. In this work, we first perform a time complexity analysis of IDS and RAPID, and show the significant advantages of the proposed method. Next we run our algorithm, along with baselines, on three public datasets. We show comparable accuracy for RAPID, with 10, 000x improvement in running time over IDS. Additionally, due to the significant improvements in running time of RAPID, we can run more extensive hyperparameter search algorithms, leading to comparable accuracy with competitive baseline models.

I. Introduction

Machine Learning (ML) models have achieved a significant position within our world, everything from loan approvals to objects detection within images are based on such models. ML and statistical models have served one of two purposes [1], to classify a data point, or explain an observation. Both explanation and classification serve important roles, but in more recent years, the role of classification or prediction has been a more dominant theme within computer science research [2]. Large black-box models have dominated these tasks. For instance, it has been noticed that the best models (with respect to classification accuracy), over hundreds of datasets are not easily interpreted [3]. While this trend towards large models has been motivated by the push to increase the prediction accuracy of these tasks, a casualty of this is the fact that humans who work with these models often fail to understand why a model took the decision that it did.