Skip to main content
  • CIS
    Members: Free
    IEEE Members: Free
    Non-members: Free
    Length: 01:24:49
19 Jul 2020

Evolutionary computation (EC) algorithms have been successfully applied to a wide range of artificial intelligence (AI) problems ranging from function optimisation, production scheduling, to evolutionary deep learning. EC researchers have been continuously developed new techniques to enhance the performance of EC algorithms. However, it is still very challenging to fully understand the behaviours of these algorithms due to the complexity of solution representations and search operators. As a result, researchers mainly rely on the performance results from experiments to suggest which algorithms perform better and to understand how novel features impact the final performance. In these studies, some questions usually left unanswered are how better results are obtained and whether the proposed algorithms behave as conceptually designed. Thus, it is critical to have an analysis tool that can help researchers gain insights on how the algorithms work and capture useful emerging patterns.
This tutorial aims at demonstrating how visualisation can be used to help researchers gain insights about EC algorithms. In this tutorial, we will review the applications of visualisation in EC such as visualising performance and generated solutions and highlight a new visualisation framework to capture high-level evolutionary patterns of EC algorithms. The following main topics will be covered in this 1.5-hour tutorial:
� Quick recap of EC algorithms and applications
� A review of visualisation techniques for EC algorithms
� AI-based visualisation (AIV) framework to reveal evolutionary patterns of EC algorithms
o Dimensionality reduction
o Topological data analysis
o Visual analytics
� Case studies:
o Evolving classifiers using genetic programming and particle swarm optimisation
o Automated design of production scheduling heuristics with genetic programming
o Evolving artificial neural networks
� Using Python to implement the AIV framework
� From AIV framework to people-centric evolutionary systems