Study of Optimisation
My university days are by now many years ago, I can hardly remember what I have learnt while I was studying for the engineering degree. Even if I can recall those subjects, most of the content has become obsolete. Older radio systems like GSM and TDMA/FDMA/CDMA are a case in point. These are 2G/3G mobile phone technologies I used to study, and we are now rolling out 5G already.
One major project I completed in my final year was the study of constrained optimisation approaches. It relates to the use of mathematics to minimise (or maximise) an objective function given a set of variables, constrains, and optimality criteria. The aim is to find a best set of variables assignment that satisfies the constraints. This can be loosely thought of as what machine learning and artificial intelligence do.
In my project, the Artificial Ant System was applied to handle the layout of graphical objects. The job was to come up with an algorithm – a step by step procedure for calculations – that can arrange a set of initially overlapping graphical objects in the most compact and elegant way (the objective function), removing the overlapping and preserving symmetry wherever possible (the constraints), using a strategy that minimises the number of moves (the optimisation).
Incidental AI-Pioneer
My work was submitted to the Pacific Rim International Conferences on Artificial Intelligence (PRICAI), and the research was accepted as one of the poster papers for AI specific application areas.
When I look at the work today, the idea is so trivial and the application of such a simple algorithm is almost laughable. I couldn’t imagine why I needed one whole year to put this together. However, in all fairness, that was more than a decade ago and I was only a student back then.
The algorithm was coded in C++ (a programing language, not the grade of my project). It’s a pity that I did not keep a copy of the program and the research paper that I wrote. Never mind, it is outdated anyway.
Fortunately, the PRICAI compiles accepted papers in every year’s conference. A 2-page summary of the paper that I co-authored with my professor can be found in the book published by Springer as lecture notes in computer science.
Although the research now looks more like a study of history than a topic in computer science, it is nonetheless some good memories to be proud of. At least I can give myself a pat on the back for being one of the pioneers in the field of artificial intelligence…
Evolving Artificial Ant Systems to Improve Layouts of Graphical Objects
Vincent Tam, Simon Koo, Kozo Sugiyama
Pages 955-956
041231_AAS