ModalImmune: Immunity Driven Unlearning via Self Destructive Training
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866917383476609024 |
|---|---|
| author | Fu, Rong Tang, WeiZhi Wang, Ziming Tan, Jia Yee Zhang, Zijian Kang, Zhaolu Qi, Muge Zhang, Shuning Fong, Simon |
| author_facet | Fu, Rong Tang, WeiZhi Wang, Ziming Tan, Jia Yee Zhang, Zijian Kang, Zhaolu Qi, Muge Zhang, Shuning Fong, Simon |
| contents | Multimodal systems are vulnerable to partial or complete loss of input channels at deployment, which undermines reliability in real-world settings. This paper presents ModalImmune, a training framework that enforces modality immunity by intentionally and controllably collapsing selected modality information during training so the model learns joint representations that are robust to destructive modality influence. The framework combines a spectrum-adaptive collapse regularizer, an information-gain guided controller for targeted interventions, curvature-aware gradient masking to stabilize destructive updates, and a certified Neumann-truncated hyper-gradient procedure for automatic meta-parameter adaptation. Empirical evaluation on standard multimodal benchmarks demonstrates that ModalImmune improves resilience to modality removal and corruption while retaining convergence stability and reconstruction capacity. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_16197 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ModalImmune: Immunity Driven Unlearning via Self Destructive Training Fu, Rong Tang, WeiZhi Wang, Ziming Tan, Jia Yee Zhang, Zijian Kang, Zhaolu Qi, Muge Zhang, Shuning Fong, Simon Machine Learning Computation and Language Multimedia Multimodal systems are vulnerable to partial or complete loss of input channels at deployment, which undermines reliability in real-world settings. This paper presents ModalImmune, a training framework that enforces modality immunity by intentionally and controllably collapsing selected modality information during training so the model learns joint representations that are robust to destructive modality influence. The framework combines a spectrum-adaptive collapse regularizer, an information-gain guided controller for targeted interventions, curvature-aware gradient masking to stabilize destructive updates, and a certified Neumann-truncated hyper-gradient procedure for automatic meta-parameter adaptation. Empirical evaluation on standard multimodal benchmarks demonstrates that ModalImmune improves resilience to modality removal and corruption while retaining convergence stability and reconstruction capacity. |
| title | ModalImmune: Immunity Driven Unlearning via Self Destructive Training |
| topic | Machine Learning Computation and Language Multimedia |
| url | https://arxiv.org/abs/2602.16197 |