Le, Huy (Monash University), Vu, Hai L. (Monash University), Nguyen, Vu (Amazon)

Multi-Objective Calibration of Large-Scale Models Using Bayesian Optimization

Scheduled for presentation during the Regular Session "S35a-Optimization, Control, and Learning for Efficient and Resilient ITS" (FR-LM-T35), Friday, November 21, 2025, 11:10−11:30, Surfers Paradise 2

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Abstract

Activity-Based Models (ABMs) are disaggregated models that simulate individual travel behaviors in detail, making them valuable tools for transportation planning and the analysis of multi-modal urban mobility. However, their calibration remains challenging due to high dimensionality, complex parameter interactions, and significant computational costs. We propose a Bayesian Optimization framework for textbf{multi-task calibration} of large-scale ABMs, using a textbf{multi-task Gaussian Process} to exploit inter-module correlations and improve sample efficiency. Calibration is structured as a two-phase process: initial per-module optimization followed by selective refinement, enhancing convergence and reducing redundant computation. We demonstrate the effectiveness of the proposed method through real-world ABM calibration, achieving improved accuracy over existing state-of-the-art methods, and validating its applicability to complex, multi-objective transportation models within advanced intelligent mobility systems.