Extracting Robust Register Automata from Neural Networks over Data Sequences

2025-11-24

AI Summary

Black-box sequential models (e.g., RNNs, Transformers) lack interpretable, formal models over continuous input domains. Method: We propose the first robustness-aware extraction framework for deterministic register automata (DRAs), integrating enhanced passive learning (L* variant) with active learning and a polynomial-time robustness checker to synthesize statistically robust and formally equivalent DRA surrogates directly from continuous-valued sequence data—without discretization assumptions. Contribution/Results: Our approach enables symbolic abstraction over continuous domains and provides formal guarantees of local robustness—either certifying robustness or generating concrete counterexamples. Experiments demonstrate stable synthesis of high-accuracy DRAs across diverse neural architectures, enabling verifiable robustness analysis of neural sequence models.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Intelligent Robots: Learning & Optimization for ROB
• Natural Language Processing: Safety and Robustness

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

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.

Problem

• Extracting interpretable automata from neural networks handling continuous data sequences
• Developing robust register automata with statistical guarantees for black-box models
• Assessing neural network robustness through automata-based certification and counterexample generation

Innovation

• Extracting deterministic register automata from black-box models
• Developing polynomial-time robustness checker for register automata
• Combining passive and active automata learning algorithms