Learning Optimal Signal Temporal Logic Decision Trees for Classification: A Max-Flow MILP Formulation

TL;DR

This paper introduces a novel method for inferring Signal Temporal Logic decision trees from data, formulated as a max-flow MILP problem, improving classification accuracy and interpretability in temporal logic-based classification tasks.

Contribution

It presents a new MILP-based framework for learning STL decision trees using max-flow algorithms, with constraint reduction techniques for better efficiency and interpretability.

Findings

Improved classification rates over previous methods

Enhanced algorithm efficiency through symmetry-based constraint reduction

Successful application to multi-class and complex STL classification scenarios

Abstract

This paper presents a novel framework for inferring timed temporal logic properties from data. The dataset comprises pairs of finite-time system traces and corresponding labels, denoting whether the traces demonstrate specific desired behaviors, e.g. whether the ship follows a safe route or not. Our proposed approach leverages decision-tree-based methods to infer Signal Temporal Logic classifiers using primitive formulae. We formulate the inference process as a mixed integer linear programming optimization problem, recursively generating constraints to determine both data classification and tree structure. Applying a max-flow algorithm on the resultant tree transforms the problem into a global optimization challenge, leading to improved classification rates compared to prior methodologies. Moreover, we introduce a technique to reduce the number of constraints by exploiting the symmetry inherent in STL primitives, which enhances the algorithm's time performance and interpretability. To assess our algorithm's effectiveness and classification performance, we conduct three case studies involving two-class, multi-class, and complex formula classification scenarios.