Artificial life based on boids model and evolutionary chaotic neural networks for creating artworks

https://doi.org/10.1016/j.swevo.2017.09.003Get rights and content

Abstract

In this paper, we propose a multi-agent based art production framework. In existing artwork creation systems, images were generated using artificial life and evolutionary computation approaches. In artificial life, swarm intelligence or Boids model, and in evolutionary computation, genetic algorithm or genetic programming are commonly used to create images. These automated artwork creation systems make it easy to create artistic images even if the users are not professional artists. Despite the high possibility of these creation systems, however, much research has not been done so far. In this paper, we propose an art production framework that generates images using multi-agents with chaotic dynamics features. Agents act on the canvas following the three rules of Boids model. In addition, each agent possesses a chaotic neural network which trained by differential evolution algorithm, so that colors can be evolved to represent a better style. As a result, we propose an art production framework for generating processing artworks that contain highly complex dynamics. Finally, we created the glitch artworks using the proposed framework, which shows a new glitch style.

Introduction

John von Neumann [50] attempted to design a self-reproducing device, which is considered to be the first study of artificial life. His work linked to the birth of cellular automata [51] and has been improved by several researchers [21], [25], [53], [7], [8]. As the public gradually begins to be interested in Jonh von Neumann's work, Langton first coined the term artificial life as an area of life study [24], [27]. Then later, Bedau defined artificial life as an interdisciplinary field that studies life and life-like processes and focuses on the essential aspects of life and the realization of life phenomena through the synthesis of these parts [1]. That is, artificial life is a field that researchers from diverse fields such as biologists, chemists, mathematicians, physicists, computer scientists, and philosophers work together.

Artificial life art is a method that creates artworks by using artificial life. These artworks contain the concepts and principles of artificial life such as nonlinearity and emergence behavior. Here, emergent behavior represents an unpredictable phenomenon that occurs only by local interactions without global information. Langton argued that although the components of artificial life and natural life are different, the concepts are same and encouraged to use them in methods related to artificial life [27]. These principles of artificial life are particularly useful for artificial life art because they can be helpful to create artworks that are hard to predict, which can give the audience greater pleasure. Similar to artificial life art, there is a method called evolutionary art that belongs to automated artwork creation systems. Evolutionary art is a method that generates images by using genetic algorithms or genetic programming [28]. Evolutionary art makes it easy to generate artistic images even if the users are not professional artists.

Despite the high possibility of these automated artwork creation systems, many studies have not been made so far. There are several problems to be solved in order to be used as more practical art creation systems. In this paper, we proposed an art production framework that possesses higher nonlinearity and emergence behavior. The artworks with these characteristics can produce patterns that are difficult to predict, which can give the audience greater pleasure. The proposed framework generates images by using multi-agents with chaotic dynamics. The agents act on a canvas following the three rules of Boids model [41]. In addition, each agent possesses a chaotic neural network, so that colors can be changed based on chaotic dynamics. As a result, we proposed an art production framework for creating artworks that contain the characteristics of artificial life art. The proposed framework can be used to create various kinds of artworks. In this paper, we use it to produce glitch artworks as an example. Glitch art is a contemporary art that is made by artificially adding errors. We confirmed that the proposed framework is possible to produce glitch artworks by adding a new glitch style to a given input image. A similar approach can be used to create other kinds of artworks.

This paper makes the following contributions. First, we proposed an art production framework that possesses higher nonlinearity and emergence behavior by integrating multi-agents with chaotic neural networks. The proposed framework combines the emergent behavior of Boids model with the complex dynamics of chaotic neural networks so that it can provide the audience the pleasure of not only finished artworks but also the process of completing artworks like processing art. Second, we used chaotic neural networks with a multilayer structure called chaotic multilayer networks, which can provide more strengthened complex dynamics than the original chaotic neural networks. That is, chaotic multilayer networks cannot only produce more complex patterns but also have higher chaotic dynamics, which can be useful to create artworks that hard to predict. Finally, we used the proposed framework to produce glitch artworks. Glitch art is one of the contemporary arts that has recently been attracting attention, and the proposed framework can generate images with a new glitch style based on a given input image.

The organization of this paper is as follows. In Chapter 2, we present related work such as artificial life art, evolutionary art, and glitch art. In Chapter 3, we introduce Boids model and chaotic neural networks, which are the background of the proposed framework. In Chapter 4, we present the proposed framework for creating artworks and its application to generate glitch art. In Chapter 5, we examine the artworks created by the proposed framework. Finally, Chapter 6 concludes this paper with the discussion.

Section snippets

Artificial life art

As a computational model for creativity was proposed using artificial intelligence [9], a method of generating designs, artworks, and music using characteristics such as the evolution of artificial life has been proposed [39], [5]. The foundation of artificial life art was laid in 1986, starting with the work of Langton, [26]. Langton uses an individual called an ant, which follows the simple rules of forwarding, rotation, and flips to create a work of art. It is the study of Tzafestas, [45]

Boids model

Reynolds proposed Boids model to implement the beauty observed in the natural world, such as bird flocking, fish schooling, and insect swarming, in the computer world [41]. Boids model is based on a bottom-up approach in which each agent interacts with only its neighbors to set the path to move. Each agent in Boids model follows three basic behavioral rules as follows:

  • 1.

    Alignment: steer for moving toward the average heading of neighbors

  • 2.

    Cohesion: steer for moving toward the mean position of

Chaotic multilayer networks

As we discussed earlier, chaotic neural networks are a model that adds chaotic dynamics to neurons to make them behave more like real neurons. The original chaotic neural networks have only one hidden layer. The reason is that chaotic neural networks have been proposed to realize the complex dynamics observed in real neurons, rather than a neural network to solve a certain problem. In this paper, we proposed chaotic neural networks with a multilayer structure called chaotic multilayer networks.

The proposed glitch art system

In this paper, we applied the proposed framework to create glitch artworks as an example. The proposed system first receives an input image and adds glitch style to it as output. Fitness is calculated using the input image as follows: First, an agent calculates its own color and the color painted at its location. At this time, if its color is similar to the input image than the currently painted color on the drawing space, the agent paints its color on the current location and its fitness will

Conclusion and future research

In this paper, we propose an art production framework with higher nonlinearity behavior. The proposed framework can create unpredictable patterns and give the audience a great pleasure. The proposed framework generates images using multi-agents with chaotic dynamics. The agent acts on the canvas according to the three rules of Boids model. In addition, each agent possesses a chaotic neural network so that it can change its own color based on chaotic mechanics. As a result, we proposed an art

Acknowledgements

This work was supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (No. NRF–2017R1C1B2012752). In addition, this work was supported by the NRF grant funded by MSIT (No. NRF–2015R1D1A1A02062017).

References (52)

  • K. Aihara et al.

    Chaotic neural networks

    Phys. Lett. A

    (1990)
  • M.A. Boden

    Creativity and artificial intelligence

    Artif. Intell.

    (1998)
  • C.G. Langton

    Self-reproduction in cellular automata

    Phys. D: Nonlinear Phenom.

    (1984)
  • C.G. Langton

    Studying artificial life with cellular automata

    Phys. D: Nonlinear Phenom.

    (1986)
  • W. Aguilar et al.

    The past, present, and future of artificial life

    Front. Robot. AI

    (2014)
  • K. Aihara et al.
  • M. Annunziato et al.

    Relazioni emergenti: experiments with the art of emergence

    Leonardo

    (2002)
  • R.F. Antunes et al.

    Two decades of evolutionary art using computational ecosystems and its potential for virtual worlds

    J. Virtual Worlds Res.

    (2014)
  • S. Aupetit, V. Bordeau, N. Monmarché, M. Slimane, G. Venturini, Interactive evolution of ant paintings. Evolutionary...
  • E.R. Banks, Information processing and transmission in cellular automata,...
  • E.R. Berlekamp, J.H. Conway, R.K. Guy, Winning ways for your mathematical plays AK Peters Natick,...
  • J. Correia et al.

    Evolving figurative images using expression-based evolutionary art

    Iccc

    (2013)
  • E. den Heijer, Evolving glitch art. International Conference on Evolutionary and Biologically Inspired Music and Art,...
  • S. Draves

    The electric sheep screen-saver: a case study in aesthetic evolution

    Evoworkshops

    (2005)
  • C.M. Fernandes

    Pherographia: drawing by ants

    Leonardo

    (2010)
  • C.M. Fernandes, A. Mora, J.J. Merelo, C. Cotta, A.C. Rosa, Photo rendering with swarms: From figurative to abstract...
  • G. Greenfield, Ant paintings using a multiple pheromone model. BRIDGES 2006 Conference Proceedings, 319-326,...
  • G. Greenfield

    Evolutionary methods for ant colony paintings

    Workshops Appl. Evolut. Comput.

    (2005)
  • G. Greenfield, On evolving multi-pheromone ant paintings. Evolutionary Computation, 2006. CEC 2006. IEEE Congress on,...
  • G. Greenfield, Ant paintings based on the seed foraging behavior of P. barbatus. 2013 BRIDGES Conference...
  • G. Greenfield. Target curves for pick-up, carry and drop mobile automata. Bridges 2014 Conference Proceedings,...
  • G. Greenfield, P. Machado, Simulating artist and critic dynamics. Proceedings of the International Joint Conference on...
  • T.J. Hutton

    Codd's self-replicating computer

    Artif. Life

    (2010)
  • C.J. Jacob et al.

    Swarmart: Interactive art from swarm intelligence

    Leonardo

    (2007)
  • T. Kowaliw et al.

    Promoting creative design in interactive evolutionary computation

    IEEE Trans. Evolut. Comput.

    (2012)
  • C. Langton, A new definition of artificial life. Retrieved...
  • Cited by (0)

    View full text