Tackling Device Data Distribution Real-time Shift via Prototype-based Parameter Editing
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866911142739181568 |
|---|---|
| author | Lv, Zheqi Zhang, Wenqiao Fu, Kairui Tian, Qi Zhang, Shengyu Su, Jiajie Chen, Jingyuan Kuang, Kun Wu, Fei |
| author_facet | Lv, Zheqi Zhang, Wenqiao Fu, Kairui Tian, Qi Zhang, Shengyu Su, Jiajie Chen, Jingyuan Kuang, Kun Wu, Fei |
| contents | The on-device real-time data distribution shift on devices challenges the generalization of lightweight on-device models. This critical issue is often overlooked in current research, which predominantly relies on data-intensive and computationally expensive fine-tuning approaches. To tackle this, we introduce Persona, a novel personalized method using a prototype-based, backpropagation-free parameter editing framework to enhance model generalization without post-deployment retraining. Persona employs a neural adapter in the cloud to generate a parameter editing matrix based on real-time device data. This matrix adeptly adapts on-device models to the prevailing data distributions, efficiently clustering them into prototype models. The prototypes are dynamically refined via the parameter editing matrix, facilitating efficient evolution. Furthermore, the integration of cross-layer knowledge transfer ensures consistent and context-aware multi-layer parameter changes and prototype assignment. Extensive experiments on vision task and recommendation task on multiple datasets confirm Persona's effectiveness and generality. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_06552 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Tackling Device Data Distribution Real-time Shift via Prototype-based Parameter Editing Lv, Zheqi Zhang, Wenqiao Fu, Kairui Tian, Qi Zhang, Shengyu Su, Jiajie Chen, Jingyuan Kuang, Kun Wu, Fei Machine Learning Computer Vision and Pattern Recognition Distributed, Parallel, and Cluster Computing Information Retrieval The on-device real-time data distribution shift on devices challenges the generalization of lightweight on-device models. This critical issue is often overlooked in current research, which predominantly relies on data-intensive and computationally expensive fine-tuning approaches. To tackle this, we introduce Persona, a novel personalized method using a prototype-based, backpropagation-free parameter editing framework to enhance model generalization without post-deployment retraining. Persona employs a neural adapter in the cloud to generate a parameter editing matrix based on real-time device data. This matrix adeptly adapts on-device models to the prevailing data distributions, efficiently clustering them into prototype models. The prototypes are dynamically refined via the parameter editing matrix, facilitating efficient evolution. Furthermore, the integration of cross-layer knowledge transfer ensures consistent and context-aware multi-layer parameter changes and prototype assignment. Extensive experiments on vision task and recommendation task on multiple datasets confirm Persona's effectiveness and generality. |
| title | Tackling Device Data Distribution Real-time Shift via Prototype-based Parameter Editing |
| topic | Machine Learning Computer Vision and Pattern Recognition Distributed, Parallel, and Cluster Computing Information Retrieval |
| url | https://arxiv.org/abs/2509.06552 |