Influence-Guided Concolic Testing of Transformer Robustness
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912612828053504 |
|---|---|
| author | Hong, Chih-Duo Wang, Yu Chang, Yao-Chen Yu, Fang |
| author_facet | Hong, Chih-Duo Wang, Yu Chang, Yao-Chen Yu, Fang |
| contents | Concolic testing for deep neural networks alternates concrete execution with constraint solving to search for inputs that flip decisions. We present an {influence-guided} concolic tester for Transformer classifiers that ranks path predicates by SHAP-based estimates of their impact on the model output. To enable SMT solving on modern architectures, we prototype a solver-compatible, pure-Python semantics for multi-head self-attention and introduce practical scheduling heuristics that temper constraint growth on deeper models. In a white-box study on compact Transformers under small $L_0$ budgets, influence guidance finds label-flip inputs more efficiently than a FIFO baseline and maintains steady progress on deeper networks. Aggregating successful attack instances with a SHAP-based critical decision path analysis reveals recurring, compact decision logic shared across attacks. These observations suggest that (i) influence signals provide a useful search bias for symbolic exploration, and (ii) solver-friendly attention semantics paired with lightweight scheduling make concolic testing feasible for contemporary Transformer models, offering potential utility for debugging and model auditing. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_23806 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Influence-Guided Concolic Testing of Transformer Robustness Hong, Chih-Duo Wang, Yu Chang, Yao-Chen Yu, Fang Software Engineering Machine Learning Concolic testing for deep neural networks alternates concrete execution with constraint solving to search for inputs that flip decisions. We present an {influence-guided} concolic tester for Transformer classifiers that ranks path predicates by SHAP-based estimates of their impact on the model output. To enable SMT solving on modern architectures, we prototype a solver-compatible, pure-Python semantics for multi-head self-attention and introduce practical scheduling heuristics that temper constraint growth on deeper models. In a white-box study on compact Transformers under small $L_0$ budgets, influence guidance finds label-flip inputs more efficiently than a FIFO baseline and maintains steady progress on deeper networks. Aggregating successful attack instances with a SHAP-based critical decision path analysis reveals recurring, compact decision logic shared across attacks. These observations suggest that (i) influence signals provide a useful search bias for symbolic exploration, and (ii) solver-friendly attention semantics paired with lightweight scheduling make concolic testing feasible for contemporary Transformer models, offering potential utility for debugging and model auditing. |
| title | Influence-Guided Concolic Testing of Transformer Robustness |
| topic | Software Engineering Machine Learning |
| url | https://arxiv.org/abs/2509.23806 |