I’ve just submitted a paper titled “Heuristic Optimization Platform For Meta And Hyper-Heuristic Solving Of Combinatorial And Continuous Problems”. As my very last piece of module work for MSc Intelligent Systems, it feels great reaching this milestone after a hard few years managing both academic and young family life.

This specific paper formed the major piece for module “Computational Intelligence Optimization”, that began with a fearful revisit of calculus, and taught me a huge amount about a building block of AI that I had not encountered before - optimisation. I really enjoyed it, and oddly enough, it was the module I was most intimidated about at the beginning.

The assignment itself was a great learning experience understanding two types of problems; continuous and combinatorial, and how to achieve “good enough” results using a variety of optimisation algorithms including particle swarm and genetic evolution.

For me, the significance of this work was taking it further to develop a hyper-heuristic, or “heuristic for meta-heuristics”, that is able to select from a pool of algorithms in real time according to the state of the problem being solved.

It proved invaluable, adding an additional repertoire of solutions on top of traditional machine learning approaches.