MuViS: Multimodal Virtual Sensing Benchmark

2026-03-13

AI Summary

This work addresses the lack of general-purpose benchmarks and transferable methodologies in virtual sensing across diverse processes, modalities, and configurations. To bridge this gap, the authors introduce MuViS—the first unified, extensible, and open-source multimodal virtual sensing benchmark—which integrates multiple datasets and standardizes preprocessing and evaluation protocols. Within this framework, the study systematically evaluates the performance of gradient-boosted decision trees alongside various deep neural network architectures. Experimental results reveal that no existing method demonstrates consistent superiority across all scenarios, underscoring the urgent need for a truly generalizable virtual sensing architecture. The MuViS platform is publicly released to foster reproducible and universally applicable research in the field.

Abstract

Virtual sensing aims to infer hard-to-measure quantities from accessible measurements and is central to perception and control in physical systems. Despite rapid progress from first-principle and hybrid models to modern data-driven methods research remains siloed, leaving no established default approach that transfers across processes, modalities, and sensing configurations. We introduce MuViS, a domain-agnostic benchmarking suite for multimodal virtual sensing that consolidates diverse datasets into a unified interface for standardized preprocessing and evaluation. Using this framework, we benchmark established approaches spanning gradient-boosted decision trees and deep neural network (NN) architectures, and show that none of these provides a universal advantage, underscoring the need for generalizable virtual sensing architectures. MuViS is released as an open-source, extensible platform for reproducible comparison and future integration of new datasets and model classes.

Problem

• virtual sensing
• multimodal
• benchmarking
• generalizable architectures
• domain-agnostic

Innovation

• multimodal virtual sensing
• benchmarking suite
• domain-agnostic
• standardized evaluation
• open-source platform