IEEE TF AADCS

As we move into the fourth industrial revolution the need for off-the-shelf machine learning tools is growing. This task force focuses on the use of evolutionary computation for automated algorithm design, configuration, and selection of machine learning techniques.

• Automated Design ranges from design of neural network architectures to determining the control flow of evolutionary algorithms and the generation of new operators and heuristics.
• Algorithm Configuration focuses on identifying the most suitable parameters and hyper-parameters to use in machine learning and search techniques.
• Algorithm Selection involves selecting an algorithm to apply to the problem at hand by mapping problem features to algorithms.

The task force aims to promote research and collaboration in this area by means of special issues of journals, special sessions and tutorials at conferences. An ultimate aim is to further research in this domain to the point where off-the-shelve tools for developing algorithms are available to the non-expert.

Focus areas of the task force include but are not limited to:

• AutoML
• Automated algorithm selection
• Algorithm portfolios
• Automated design of operators
• Automated hybridization of algorithms
• Automated parameter configuration and adaptation
• Automated hyper-parameter selection
• Automated feature selection
• Automated model selection
• Automated heuristic generation
• Automated operator creation
• Hyper-heuristics
• Multilevel metaheuristics
• Reactive search
• Self-configuration
• Self-adaption
• Neuroevolution
• Theoretical aspects