Verifying Properties of Binary Neural Networks Using Sparse Polynomial Optimization
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_ | 1866912327404617728 |
|---|---|
| author | Yang, Jianting Ðurašinović, Srećko Lasserre, Jean-Bernard Magron, Victor Zhao, Jun |
| author_facet | Yang, Jianting Ðurašinović, Srećko Lasserre, Jean-Bernard Magron, Victor Zhao, Jun |
| contents | This paper explores methods for verifying the properties of Binary Neural Networks (BNNs), focusing on robustness against adversarial attacks. Despite their lower computational and memory needs, BNNs, like their full-precision counterparts, are also sensitive to input perturbations. Established methods for solving this problem are predominantly based on Satisfiability Modulo Theories and Mixed-Integer Linear Programming techniques, which are characterized by NP complexity and often face scalability issues.
We introduce an alternative approach using Semidefinite Programming relaxations derived from sparse Polynomial Optimization. Our approach, compatible with continuous input space, not only mitigates numerical issues associated with floating-point calculations but also enhances verification scalability through the strategic use of tighter first-order semidefinite relaxations. We demonstrate the effectiveness of our method in verifying robustness against both $\|.\|_\infty$ and $\|.\|_2$-based adversarial attacks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_17049 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Verifying Properties of Binary Neural Networks Using Sparse Polynomial Optimization Yang, Jianting Ðurašinović, Srećko Lasserre, Jean-Bernard Magron, Victor Zhao, Jun Machine Learning Optimization and Control This paper explores methods for verifying the properties of Binary Neural Networks (BNNs), focusing on robustness against adversarial attacks. Despite their lower computational and memory needs, BNNs, like their full-precision counterparts, are also sensitive to input perturbations. Established methods for solving this problem are predominantly based on Satisfiability Modulo Theories and Mixed-Integer Linear Programming techniques, which are characterized by NP complexity and often face scalability issues. We introduce an alternative approach using Semidefinite Programming relaxations derived from sparse Polynomial Optimization. Our approach, compatible with continuous input space, not only mitigates numerical issues associated with floating-point calculations but also enhances verification scalability through the strategic use of tighter first-order semidefinite relaxations. We demonstrate the effectiveness of our method in verifying robustness against both $\|.\|_\infty$ and $\|.\|_2$-based adversarial attacks. |
| title | Verifying Properties of Binary Neural Networks Using Sparse Polynomial Optimization |
| topic | Machine Learning Optimization and Control |
| url | https://arxiv.org/abs/2405.17049 |