RIE-SenseNet: Riemannian Manifold Embedding of Multi-Source Industrial Sensor Signals for Robust Pattern Recognition

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Wang, Xu, Han, Puyu, Kang, Jiaju, Pan, Weichao, Gong, Luqi
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910934159589376
author Wang, Xu
Han, Puyu
Kang, Jiaju
Pan, Weichao
Gong, Luqi
author_facet Wang, Xu
Han, Puyu
Kang, Jiaju
Pan, Weichao
Gong, Luqi
contents Industrial sensor networks produce complex signals with nonlinear structure and shifting distributions. We propose RIE-SenseNet, a novel geometry-aware Transformer model that embeds sensor data in a Riemannian manifold to tackle these challenges. By leveraging hyperbolic geometry for sequence modeling and introducing a manifold-based augmentation technique, RIE-SenseNet preserves sensor signal structure and generates realistic synthetic samples. Experiments show RIE-SenseNet achieves >90% F1-score, far surpassing CNN and Transformer baselines. These results illustrate the benefit of combining non-Euclidean feature representations with geometry-consistent data augmentation for robust pattern recognition in industrial sensing.
format Preprint
id arxiv_https___arxiv_org_abs_2502_02428
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RIE-SenseNet: Riemannian Manifold Embedding of Multi-Source Industrial Sensor Signals for Robust Pattern Recognition
Wang, Xu
Han, Puyu
Kang, Jiaju
Pan, Weichao
Gong, Luqi
Machine Learning
Industrial sensor networks produce complex signals with nonlinear structure and shifting distributions. We propose RIE-SenseNet, a novel geometry-aware Transformer model that embeds sensor data in a Riemannian manifold to tackle these challenges. By leveraging hyperbolic geometry for sequence modeling and introducing a manifold-based augmentation technique, RIE-SenseNet preserves sensor signal structure and generates realistic synthetic samples. Experiments show RIE-SenseNet achieves >90% F1-score, far surpassing CNN and Transformer baselines. These results illustrate the benefit of combining non-Euclidean feature representations with geometry-consistent data augmentation for robust pattern recognition in industrial sensing.
title RIE-SenseNet: Riemannian Manifold Embedding of Multi-Source Industrial Sensor Signals for Robust Pattern Recognition
topic Machine Learning
url https://arxiv.org/abs/2502.02428