DeepDFA: Automata Learning through Neural Probabilistic Relaxations

TL;DR

DeepDFA introduces a neural probabilistic approach to automata learning that improves scalability, noise resilience, and interpretability over traditional methods by combining probabilistic relaxations with neural network techniques.

Contribution

It presents a novel differentiable model for DFA learning that outperforms existing combinatorial methods in scalability and noise robustness.

Findings

Accurate and fast automata learning on complex languages

Enhanced robustness to noisy input and labels

Improved interpretability of learned automata

Abstract

In this work, we introduce DeepDFA, a novel approach to identifying Deterministic Finite Automata (DFAs) from traces, harnessing a differentiable yet discrete model. Inspired by both the probabilistic relaxation of DFAs and Recurrent Neural Networks (RNNs), our model offers interpretability post-training, alongside reduced complexity and enhanced training efficiency compared to traditional RNNs. Moreover, by leveraging gradient-based optimization, our method surpasses combinatorial approaches in both scalability and noise resilience. Validation experiments conducted on target regular languages of varying size and complexity demonstrate that our approach is accurate, fast, and robust to noise in both the input symbols and the output labels of training data, integrating the strengths of both logical grammar induction and deep learning.