Extracting Robust Register Automata from Neural Networks over Data Sequences
Hong, Chih-Duo, Jiang, Hongjian, Lin, Anthony W., Markgraf, Oliver, Parsert, Julian, Tan, Tony
–arXiv.org Artificial Intelligence
Automata extraction is a method for synthesising interpretable surrogates for black-box neural models that can be analysed symbolically. Existing techniques assume a finite input alphabet, and thus are not directly applicable to data sequences drawn from continuous domains. We address this challenge with deterministic register automata (DRAs), which extend finite automata with registers that store and compare numeric values. Our main contribution is a framework for robust DRA extraction from black-box models: we develop a polynomial-time robustness checker for DRAs with a fixed number of registers, and combine it with passive and active automata learning algorithms. This combination yields surrogate DRAs with statistical robustness and equivalence guarantees. As a key application, we use the extracted automata to assess the robustness of neural networks: for a given sequence and distance metric, the DRA either certifies local robustness or produces a concrete counterexample. Experiments on recurrent neural networks and transformer architectures show that our framework reliably learns accurate automata and enables principled robustness evaluation. Overall, our results demonstrate that robust DRA extraction effectively bridges neural network interpretability and formal reasoning without requiring white-box access to the underlying network.
arXiv.org Artificial Intelligence
Nov-25-2025
- Country:
- Asia
- Europe
- Estonia > Harju County
- Tallinn (0.04)
- United Kingdom
- England
- Cambridgeshire > Cambridge (0.04)
- Merseyside > Liverpool (0.04)
- North Sea > Central North Sea (0.04)
- England
- France > Île-de-France
- Slovenia > Drava
- Municipality of Benedikt > Benedikt (0.04)
- Middle East > Cyprus
- Switzerland (0.04)
- Spain
- Germany
- North Rhine-Westphalia > Cologne Region
- Aachen (0.04)
- Rhineland-Palatinate
- Kaiserslautern (0.05)
- Landau (0.04)
- North Rhine-Westphalia > Cologne Region
- Austria (0.04)
- Netherlands > North Holland
- Amsterdam (0.04)
- Sweden > Stockholm
- Stockholm (0.04)
- Estonia > Harju County
- North America
- Canada
- Alberta > Census Division No. 6
- Calgary Metropolitan Region > Calgary (0.04)
- Quebec > Montreal (0.04)
- Alberta > Census Division No. 6
- United States
- California > San Francisco County
- San Francisco (0.14)
- Minnesota > Hennepin County
- Minneapolis (0.14)
- New York
- Bronx County > New York City (0.04)
- Kings County > New York City (0.04)
- New York County > New York City (0.14)
- Queens County > New York City (0.04)
- Richmond County > New York City (0.04)
- Pennsylvania > Allegheny County
- Pittsburgh (0.04)
- Texas > Travis County
- Austin (0.04)
- Washington > King County
- Seattle (0.04)
- California > San Francisco County
- Canada
- South America > Argentina
- Pampas > Buenos Aires F.D. > Buenos Aires (0.04)
- Genre:
- Research Report > New Finding (1.00)
- Industry:
- Banking & Finance (0.45)
- Technology: