Department of Computer Science
Brock University

Research

Some of the research areas studied by BICIG members include...

• combinatorial optimization
• complex network modeling
• multi-objective optimization
• bio-network synthesis
• evolutionary design

BICIG research uses many computational intelligence techniques, such as...

• genetic algorithms
• genetic programming
• particle swarm optimization
• ant colony optimization
• neural networks

The following describes a selection of research topics explored by BICIG members.

• Combinatorial optimization of static systems.

  Example papers:

• Combinatorial optimization of dynamic systems.

  Example papers:

• Modelling of complex networks.

  Example papers:

• Multi-objective optimization and feature reduction. Complex real-world problems often involve multiple criteria for evaluating success. These objectives often interact with each other in complex and conflicting ways. It is especially challenging when there are a large number of competing objectives. Related to this, classification problems often involve a high number of features, which can make classification accuracy difficult. BICIG research has explored issues in multi-objective optimization and feature reduction, in order to make computational algorithms more effective.

  Example papers:

  • A. Awuley and B.J. Ross. Feature Selection and Classification Using Age Layered Population Structure Genetic Programming. (Submitted for publication.)

• Bio-network modelling. Complex biological systems can be modeled and simulated with formal systems. One example of such a formal system is a stochastic process algebra. Creating formal models for complex behaviours is challenging. Research is ongoing in using genetic programming to automatically synthesize bio-network models written in different stochastic process algebra languages. Work has examined the stochastic pi-calculus, a gene gate regulatory language, and the PIM bio-modeling language.

  Example papers:

  • B.J. Ross. Using Multi-objective Genetic Programming to Evolve Stochastic Logic Gate Circuits. CIBCB 2015, Niagara Falls, Canada.
  • B.J. Ross. The Evolution of Higher-Level Biochemical Reaction Models. Genetic Programming and Evolvable Machines, v.13, n.1, 2012, pp. 3-31.
  • J. Imada and B.J. Ross. Evolutionary Synthesis of Stochastic Gene Network Models using Feature-based Search Spaces. New Generation Computing, v.29, n.4, October 2011, pp. 365-390.

• Architecture, 3D modelling, and evolutionary design. Research is using genetic programming in architecture and 3D modelling. Genetic programming has been used to produce 3D architectures using split grammars and L-systems. The evolution of energy efficient building models has been studied. Aethetic 3D models were evolved with genetic programming, using Ralph's aesthetic model. Floor plans complying to user specifications were evolved with a multi-objective genetic algorithm. Interior illumination for 3D spaces was also examined.

  Example papers:

  • K. Moylan and B.J. Ross. Interior Illumination Design Using Genetic Programming. EvoMusArt 2015, Copenhagen, Denmark, April 2015. pp.148-160. Best paper nomination.
  • M.M. Oraei Gholami and Brian J. Ross. Passive Solar Building Design Using Genetic Programming, GECCO 2014, Vancouver, July 2014.
  • S. Bergen and B.J. Ross. Aesthetic 3D Model Evolution. Genetic Programming and Evolvable Machines, April 2013. (Best paper award at EvoMusArt 2012).

• 2D graphics and evolutionary art. Research is exploring ways to use evolutionary algorithms to generate artistic images. Some work has used Ralph's model of aesthetics as a fitness criteria, in order to define a mathematical model of image quality for automated evolution. Work has explored aesthetic procedural textures, non-photorealistic image generation as seen in natural media such as water colour and oil paints, and vector graphics. Particle swarms were used for automated photography in a virtual 3D environment.

  Example papers:

  • M. Baniasadi and B.J. Ross. Exploring Non-photorealistic Rendering with Genetic Programming. Genetic Programming and Evolvable Machines, 16(2), June 2015, pp. 211-239, DOI 10.1007/s10710-014-9234-0. (Winner of Evolutionary Art Competition, GECCO 2014).
  • W. Barry and B.J. Ross. Virtual Photography Using Multi-objective Particle Swarm Optimization, GECCO 2014, Vancouver, July 2014.
  • C. Neufeld, B.J. Ross, W. Ralph. The Evolution of Artistic Filters, In The Art of Artificial Evolution, J. Romero and P. Machado (eds.), pp. 335-356, Springer, 2008.