# Learning deterministic finite automata with a smart state labeling evolutionary algorithm

@article{Lucas2005LearningDF, title={Learning deterministic finite automata with a smart state labeling evolutionary algorithm}, author={Simon M. M. Lucas and T. Jeff Reynolds}, journal={IEEE Transactions on Pattern Analysis and Machine Intelligence}, year={2005}, volume={27}, pages={1063-1074} }

Learning a deterministic finite automaton (DFA) from a training set of labeled strings is a hard task that has been much studied within the machine learning community. It is equivalent to learning a regular language by example and has applications in language modeling. In this paper, we describe a novel evolutionary method for learning DFA that evolves only the transition matrix and uses a simple deterministic procedure to optimally assign state labels. We compare its performance with the…

## Figures and Tables from this paper

## 83 Citations

Active Coevolutionary Learning of Deterministic Finite Automata

- Computer ScienceJ. Mach. Learn. Res.
- 2005

It is shown that this algorithm outperforms one version of the most powerful set of algorithms for grammatical inference, evidence driven state merging (EDSM), on randomly-generated DFAs, and is proposed as the estimation-exploration algorithm (EEA).

Learning DFAs by Evolving Short Sequences of Merges

- Computer ScienceICGI
- 2021

This paper implemented an evolutionary algorithm in which the chromosomes encode short sequences of merges selected from a subset of high statereduction merges and uses the state-of-the-art EDSM algorithm as a baseline to compare results to and observe that it can find low-error hypotheses or the exact target DFAs with a considerably higher likelihood.

An Incremental-Evolutionary Approach for Learning Deterministic Finite Automata

- Computer Science2006 IEEE International Conference on Evolutionary Computation
- 2006

This work proposes an approach for learning deterministic finite automata (DFA) that combines incremental learning and evolutionary algorithms and shows that this approach performs well regardless the level of noise present in the training set.

Learning Finite-State Transducers: Evolution Versus Heuristic State Merging

- Computer ScienceIEEE Transactions on Evolutionary Computation
- 2007

For certain classes of FST induction problem, the RMHC offers the best performance of any known algorithm, in order of increasing computational complexity: string equality, Hamming distance, and edit distance.

Active Learning of Nondeterministic Finite State Machines

- Computer Science
- 2013

It is proved that the refined ONFSM learning algorithm can infer the corresponding ONFSMs of the unknown systems when the number of tries for the same query is adequate to guarantee the complete testing assumption.

A more bio-plausible approach to the evolutionary inference of finite state machines

- Computer ScienceGECCO '07
- 2007

An evolutionary algorithm with a novel representation and a new fitness function for inference of Moore finite-state machines of limited size is proposed and compared with one of the latest evolutionary techniques.

Genetic algorithm for induction of finite automata with continuous and discrete output actions

- Computer ScienceGECCO
- 2011

A genetic algorithm for induction of finite automata with continuous and discrete output actions, based on equations system solving, which aims to maximize the value of fitness function.

A statistically aligned recombination operator for finite state machines

- Computer Science2007 IEEE Congress on Evolutionary Computation
- 2007

Results show that when in the neighbourhood of the target machine, the aligned crossover operator reaches the optimum in far few steps than either a naive crossover operator or a standard flip-style mutation operator.

Language Inference with Multi-head Automata through Reinforcement Learning

- Computer Science2020 International Joint Conference on Neural Networks (IJCNN)
- 2020

The purpose of this paper is to use reinforcement learning to model learning agents which can recognize formal languages and shows that genetic algorithm performs better than Q- learning algorithm in general but Q-learning algorithm finds solutions faster for regular languages.

A Survey on Inductive Learning

- Computer Science
- 2005

This report surveys some of the key results in automata learning, grammar induction, hidden markov model learning and symbolic statistical modeling.

## References

SHOWING 1-10 OF 44 REFERENCES

Learning DFA: evolution versus evidence driven state merging

- Computer ScienceThe 2003 Congress on Evolutionary Computation, 2003. CEC '03.
- 2003

A new method for evolving DFAs, where only the transition matrix is evolved, and the state labels are chosen to optimize the fit between final states and training set labels, which reduces the size and in particular, the complexity of the search space.

BLUE∗: a Blue-Fringe Procedure for Learning DFA with Noisy Data

- Computer Science

Rpni and edsm are exact learning algorithms because they fit the data: it is proven that if E contains some special (characteristic) strings, then these algorithms are able to infer, in polynomial time, the dfa that produced the data, called the target dfa.

On State Merging in Grammatical Inference: A Statistical Approach for Dealing with Noisy Data

- Computer ScienceICML
- 2003

This paper proposes a statistical approach for dealing with noisy data during the inference of automata, by the state merging algorithm RPNI, based on a proportion comparison test, which relaxes the merging rule of RPNi without endangering the generalization error.

An Incremental Learning Algorithm for Deterministic Finite Automata using Evolutionary Algorithms

- Computer Science
- 2004

This work proposes an approach for learning Deterministic Finite Automata (DFA) that combines Incremental Learning and Evolutionary Algorithms and the Hybrid Adaptive evolutionary Algorithm proposed by Gomez is used.

A Genetic Algorithm for the Induction of Nondeterministic Pushdown Automata

- Computer Science
- 1995

A genetic algorithm used to infer pushdown automata from legal and illegal examples of a language is presented, and an introduction into grammatical inference is given, and related work ingrammatical inference using genetic algorithms is discussed.

Evolving Finite State Transducers: Some Initial Explorations

- Computer ScienceEuroGP
- 2003

This paper considers three string-distance based fitness functions, compute their fitness distance correlations, and presents results on using two of these (Strict and Hamming) to evolve FSTs.

Beyond EDSM

- Computer ScienceICGI
- 2002

A small modification to W-EDSM is made which improves the performance of the algorithm by 27% and is suggested to suggest other avenues for futher enhancement.

A Sampling-Based Heuristic for Tree Search Applied to Grammar Induction

- Computer ScienceAAAI/IAAI
- 1998

The aim of this paper is to describe the underlying heuristics used by SAGE to address problems belonging to that class of problems for which it is difficult to design transformation operators to perform local search because of intrinsic constraints in the definition of the problem itself.

Efficient search techniques for the inference of minimum size finite automata

- Computer ScienceProceedings. String Processing and Information Retrieval: A South American Symposium (Cat. No.98EX207)
- 1998

The application of a set of techniques known as dependency directed backtracking to the inference of the minimum size deterministic automaton consistent with a prespecified set of input/output strings yields an algorithm that is orders of magnitude faster than existing approaches.

Cryptographic Limitations on Learning Boolean Formulae and Finite Automata

- Computer Science, MathematicsMachine Learning: From Theory to Applications
- 1993

It is proved that a polynomial-time learning algorithm for Boolean formulae, deterministic finite automata or constant-depth threshold circuits would have dramatic consequences for cryptography and number theory and is applied to obtain strong intractability results for approximating a generalization of graph coloring.