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

Authors:
Anton Alexandrov;Alexey Sergushichev;Sergey Kazakov;Fedor Tsarev
Affiliations:
St. Petersburg State University of IT, Mechanics and Optics, St. Petersburg, Russian Fed.;St. Petersburg State University of IT, Mechanics and Optics, St. Petersburg, Russian Fed.;St. Petersburg State University of IT, Mechanics and Optics, St. Petersburg, Russian Fed.;St. Petersburg State University of IT, Mechanics and Optics, St. Petersburg, Russian Fed.
Venue:
Proceedings of the 13th annual conference companion on Genetic and evolutionary computation
Year:
2011

Citing 6
Cited 2

Genetic programming: on the programming of computers by means of natural selection

Genetic programming: on the programming of computers by means of natural selection
Logic Control and “Reactive” Systems: Algorithmization and Programming

Automation and Remote Control
Learning Deterministic Finite Automata with a Smart State Labeling Evolutionary Algorithm

IEEE Transactions on Pattern Analysis and Machine Intelligence
Learning for control from multiple demonstrations

Proceedings of the 25th international conference on Machine learning
Artificial Intelligence: A Modern Approach

Artificial Intelligence: A Modern Approach
Evolving finite state transducers: some initial explorations

EuroGP'03 Proceedings of the 6th European conference on Genetic programming

MuACOsm: a new mutation-based ant colony optimization algorithm for learning finite-state machines

Proceedings of the 15th annual conference on Genetic and evolutionary computation
Search-based construction of finite-state machines with real-valued actions: new representation model

Proceedings of the 15th annual conference companion on Genetic and evolutionary computation

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we describe a genetic algorithm for induction of finite automata with continuous and discrete output actions. Input data for the algorithm is a set of tests. Each test consists of two sequences: input events and output actions. In previous works output actions were discrete, i.e. selected from the finite set, in this work output actions can also be continuous, i.e. represented by real numbers. Only the structure of automaton transitions graph is evolved by the genetic algorithm. Values of output actions are found using transition labeling algorithm, which aim is to maximize the value of fitness function. New transition labeling algorithm is proposed. It also works with continuous output actions and is based on equations system solving. In case of proper selection of fitness function, equations in this system are linear and it can be solved by the Gaussian elimination method. The unmanned airplane performing the loop is considered as an example of the controlled object.