Computer Science > Computational Engineering, Finance, and Science
arXiv:2512.09319 (cs)
Abstract:Industrial cyber physical systems operate under heterogeneous sensing, stochastic dynamics, and shifting process conditions, producing data that are often incomplete, unlabeled, imbalanced, and domain shifted. High-fidelity datasets remain costly, confidential, and slow to obtain, while edge devices face strict limits on latency, bandwidth, and energy. These factors restrict the practicality of centralized deep learning, hinder the development of reliable digital twins, and increase the risk of error escape in safety-critical applications. Motivated by these challenges, this dissertation develops an efficiency grounded computation…
Computer Science > Computational Engineering, Finance, and Science
arXiv:2512.09319 (cs)
Abstract:Industrial cyber physical systems operate under heterogeneous sensing, stochastic dynamics, and shifting process conditions, producing data that are often incomplete, unlabeled, imbalanced, and domain shifted. High-fidelity datasets remain costly, confidential, and slow to obtain, while edge devices face strict limits on latency, bandwidth, and energy. These factors restrict the practicality of centralized deep learning, hinder the development of reliable digital twins, and increase the risk of error escape in safety-critical applications. Motivated by these challenges, this dissertation develops an efficiency grounded computational framework that enables data lean, physics-aware, and deployment ready intelligence for modern manufacturing environments. The research advances methods that collectively address core bottlenecks across multimodal and multiscale industrial scenarios. Generative strategies mitigate data scarcity and imbalance, while semi-supervised learning integrates unlabeled information to reduce annotation and simulation demands. Physics-informed representation learning strengthens interpretability and improves condition monitoring under small-data regimes. Spatially aware graph-based surrogate modeling provides efficient approximation of complex processes, and an edge cloud collaborative compression scheme supports real-time signal analytics under resource constraints. The dissertation also extends visual understanding through zero-shot vision language reasoning augmented by domain specific retrieval, enabling generalizable assessment in previously unseen scenarios. Together, these developments establish a unified paradigm of data efficient and resource aware intelligence that bridges laboratory learning with industrial deployment, supporting reliable decision-making across diverse manufacturing systems.
| Comments: | 2025, University of Connecticut |
| Subjects: | Computational Engineering, Finance, and Science (cs.CE); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2512.09319 [cs.CE] |
| (or arXiv:2512.09319v1 [cs.CE] for this version) | |
| https://doi.org/10.48550/arXiv.2512.09319 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Qianyu Zhou [view email] [v1] Wed, 10 Dec 2025 05:08:55 UTC (21,411 KB)