Wi-Fi-based Personnel Identity Recognition: Addressing Dataset Imbalance with C-DDPMs
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913342278336512 |
|---|---|
| author | Bian, Jichen Tan, Chong Tang, Peiyao Zheng, Min |
| author_facet | Bian, Jichen Tan, Chong Tang, Peiyao Zheng, Min |
| contents | Wireless sensing technologies become increasingly prevalent due to the ubiquitous nature of wireless signals and their inherent privacy-friendly characteristics. Device-free personnel identity recognition, a prevalent application in wireless sensing, is susceptibly challenged by imbalanced channel state information (CSI) datasets. This letter proposes a novel method for CSI dataset augmentation that employs Conditional Denoising Diffusion Probabilistic Models (C-DDPMs) to generate additional samples that address class imbalance issues. The augmentation markedly improves classification accuracies on our homemade dataset, elevating all classes to above 94%. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_04829 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Wi-Fi-based Personnel Identity Recognition: Addressing Dataset Imbalance with C-DDPMs Bian, Jichen Tan, Chong Tang, Peiyao Zheng, Min Signal Processing Wireless sensing technologies become increasingly prevalent due to the ubiquitous nature of wireless signals and their inherent privacy-friendly characteristics. Device-free personnel identity recognition, a prevalent application in wireless sensing, is susceptibly challenged by imbalanced channel state information (CSI) datasets. This letter proposes a novel method for CSI dataset augmentation that employs Conditional Denoising Diffusion Probabilistic Models (C-DDPMs) to generate additional samples that address class imbalance issues. The augmentation markedly improves classification accuracies on our homemade dataset, elevating all classes to above 94%. |
| title | Wi-Fi-based Personnel Identity Recognition: Addressing Dataset Imbalance with C-DDPMs |
| topic | Signal Processing |
| url | https://arxiv.org/abs/2404.04829 |