Utilising computer science to save businesses time and money
Research undertaken by Stirling’s computer science researchers has provided invaluable efficiency solutions for large-scale organisations managing expansive workforces and complex operations.
By increasing the speed at which essential software runs and improving the efficiency of jobs being allocated to team members, academics have allowed staff to make better use of their time, contributing to vast economic savings.
Largely funded by the Engineering and Physical Sciences Research Council and led by Dr Sandy Brownlee, Stirling research has included investigations into computational modelling, metaheuristic design and search techniques. Their approaches have been adopted within several global companies and led to numerous industrial collaborations.
Since its launch in early 2018 to the end of 2020, iRoster has saved British Telecommunications (BT) over £1 million.
Improving airline scheduling
KLM Royal Dutch Airlines is an international airline carrying more than 34 million passengers per year to 162 destinations.
Dr Brownlee worked with the airline to improve efficiency across the board by better utilising their in-house application called Opium, which tests the schedules for all KLM flights worldwide. Stirling’s tuning and optimisation of software meant Opium was able to make more accurate predictions and run 20% faster.
By successfully automatically improving computer code, the reliability of scheduling for 200,000 KLM flights increased significantly, reaching around 60 million passengers. Not only did this reduce the airline’s costs, it also saved a total of two years of personnel time.
As a result, the automated software tuning was adopted across the KLM and Air France development teams. The improved software was in place for two years and led to a fundamental shift in best practice at the company, which was geared toward the setting of software parameters, saving developer time and vastly improving efficiencies.
Improvements to KLM's in-house application significantly increased the reliability of scheduling for 200,000 flights, benefiting around 60 million passengers and saving two years of personnel time.
Improving patient management
Computer scientists have also applied their optimisation work in healthcare settings, working with the Janus Rehabilitation Centre in Iceland.
The independent vocational rehabilitation centre provides a structured programme of therapy to help people recovering from physical and mental health issues, helping them back into education or work.
The centre, which treats around 150 patients at any one time, is one of the three largest facilities of its kind in the country. By helping people return to employment or education, the centre saved Icelandic society £8m between 2017 and 2019.
To help contribute to further savings and help things run as smoothly as possible, Stirling researchers created bespoke, automated bug-fixing software for the company. This freed up time for the Janus team to improve functionality and performance elsewhere.
By giving the centre access to predictive modelling software tailored to their specific needs, and providing ongoing support and maintenance at a fraction of the usual cost, researchers helped the centre reduce their development costs. In turn, the centre saved the Icelandic economy a staggering sum of at least £4.2m.
Stirling’s Dr Saemundur Haraldsson also developed a predictive modelling tool for Janus which has been used in the decision-making process for 395 patients since January 2017. This led to a 39% increase in the success rate of people returning to education or work, which translates to an extra £1.4m in savings to Icelandic society.
The work has led to ongoing collaborations between the Stirling research team, Linköping University, Karlstad University, an Icelandic vocational rehabilitation centre, the Icelandic Government, a Swedish Coordination Association (Samordningsförbund), and two Swedish local government councils.
AI, and indeed most software, is still largely programmed by humans. Yet there are signs that this might be changing, as several programming tools are emerging which help to automate software testing.
Over the past couple of decades, the research literature has filled up with endless new nature-based metaphors for algorithms. But is it possible to develop one tool that can solve complex problems better than all the others?
In this talk, Dr Sandy Brownlee looks at a few approaches taken to use visualisation for trying to improve trust in the solutions of search-based optimisation.
Haraldsson S, Brynjolfsdottir RD, Gudnason V, Tomasson K & Siggeirsdottir K (2018) Predicting Changes in Quality of Life for Patients in Vocational Rehabilitation. In: EAIS 2018 proceedings. Evolving and Adaptive Intelligent Systems, Rhodes, Greece, 25.05.2018-27.05.2018. Piscataway, NJ, USA: IEEE. https://doi.org/10.1109/EAIS.2018.8397182
Wang X, Brownlee AEI, Woodward JR, Weiszer M, Mahfouf M & Chen J (2021) Aircraft taxi time prediction: Feature importance and their implications. Transportation Research Part C: Emerging Technologies, 124, Art. No.: 102892. https://doi.org/10.1016/j.trc.2020.102892
Brownlee A, Burles N & Swan J (2017) Search-based energy optimization of some ubiquitous algorithms. IEEE Transactions on Emerging Topics in Computational Intelligence, 1 (3), pp. 188-201. https://doi.org/10.1109/TETCI.2017.2699193
Reid, K. N., Li, J., Brownlee, A. E. I., Kern, M., Veerapen, N., Swan, J. & Owusu, G. A Hybrid Metaheuristic Approach to a Real World Employee Scheduling Problem. Proc. of the Genetic and Evolutionary Computation Conference 2019, Prague, Czech Republic, pp 1311-1318
Brownlee, A.E.I., Adair, J., Haraldsson, S.O., and Jabbo, J. Exploring the Accuracy - Energy Trade-off in Machine Learning. Proceedings of the Genetic Improvement Workshop, International Conference on Software Engineering 2021. Madrid, Spain, pp 11-18.
Haraldsson, S.O., Woodward, J.R., Brownlee, A. E. I. & Siggeirsdottir, K. Fixing Bugs in Your Sleep: How Genetic Improvement Became an Overnight Success. Genetic Improvement Workshop in: Companion Proc. of the Genetic and Evolutionary Computation Conference 2017, Berlin, Germany, pp 1513-1520