7 March 2019. Challenging an algorithm is not for the fainthearted, and requires rigorous testing to prove its viability and value.
University of Canberra Assistant Professor of Engineering Dr Shahid Hussain is all too familiar with the arduous process of developing an algorithm.
Dr Hussain has invented a new method for Multi Objective Optimisation called the Equitable Fuzzy Sorting Genetic Algorithm (EFSGA), which contested well-established algorithms NSGA-II and III.
Multi objective optimisation is also referred to as multi-objective programming; it’s a decision-making tool that resolves mathematical optimisation problems with more than one objective.
The NSGA-II and III algorithms were proposed by Professor Kalyanomoy Deb of Michigan State University. The algorithm has been cited over 18,300 times on Scopus and more than 29,500 times on Google Scholar, making it a highly-recognised multi objective genetic algorithm.
The development of EFSGA has been seven years in the making. While completing his PhD at the University of Auckland in 2012, Dr Hussain realised that the NSGA-II and III algorithm was not providing him with the results he needed to further his research in engineering solutions. He started working on an alternative algorithm with few fellow students, and today his efforts are paying off.
The task of contesting an algorithm requires robust testing and evaluation to determine effectiveness across a broad range of applications.
“When introducing a new algorithm, benchmark standards need to be met. We tested the new algorithm against benchmark algorithms and it performed extremely well. This demonstrated its effectiveness in multi-field optimisation,” said Dr Hussain.
Dr Hussain has used the algorithm in the initial development stage of a robotic surgical tool that can move in six orientations, as well as a robot that can provide physiotherapy solutions for patients with wrist injuries. These are still under development but have proved themselves effective using the EFSGA.
Dr Hussain recently published a paper on the proposed algorithm in IEEE journal, supporting his argument that the algorithm is providing solutions to multiple tasks in engineering fields.
It has further proven itself as a multifaceted solution to commercial, medical and engineering sectors, having been applied across Faculties at the University of Canberra in biomedical health areas such as breast cancer formulations, supply chain management, and the resolution of informatics problems.
“We are already working on a new version of EFSGA, to incorporate user preferences and to validate its effectiveness on robotic mechanisms,” Dr Hussain said.
“We are trying to make it more robust and suitable for the community by solving real world problems, which will have a direct impact on daily life.”
