WCDT: Systematic WCET Optimization for Decision Tree Implementations
Abstract
Machine-learning models are increasingly deployed on resource-constrained embedded systems with strict timing constraints. In such scenarios, the worst-case execution time (WCET) of the models is required to ensure safe operation. Specifically, decision trees are a prominent class of machine-learning models and the main building blocks of tree-based ensemble models (e.g., random forests), which are commonly employed in resource-constrained embedded systems. In this paper, we develop a systematic approach for WCET optimization of decision tree implementations. To this end, we introduce a linear surrogate model that estimates the execution time of individual paths through a decision tree based on the path's length and the number of taken branches. We provide an optimization algorithm that constructively builds a WCET-optimal implementation of a given decision tree with respect to this surrogate model. We experimentally evaluate both the surrogate model and the WCET-optimization algorithm. The evaluation shows that the optimization algorithm improves analytically determined WCET by up to compared to an unoptimized implementation.
Cite
@article{arxiv.2501.17428,
title = {WCDT: Systematic WCET Optimization for Decision Tree Implementations},
author = {Nils Hölscher and Christian Hakert and Georg von der Brüggen and Jian-Jia Chen and Kuan-Hsun Chen and Jan Reineke},
journal= {arXiv preprint arXiv:2501.17428},
year = {2025}
}