The Annual Humies prize is awarded at the International Conference on Genetic and Evolutionary Computation for human-competitive results that have been produced by any form of genetic and evolutionary computation.
Working with colleagues Prof. Barry Gardiner (INRA, France) and Dr Kana Kamimura (Shinshu, Japan) , Prof. Hart and Dr Sim developed a novel method to improve predictive models to identify trees likely to be damaged in major storms: these models help forest planners develop policies for forest management that minimise wind-damage risk.
Prof. Hart and Dr Sim were awarded the Bronze Medal in Kyoto, Japan for work on improving computational models to predict damage to individual trees in forests during a storm
Their approach used a technique called Genetic Programming to evolve an expanded feature set that was combined with an existing classifier. When tested on data obtained following catastrophic storms in the South-West of France, the new model was able to predict damage to individual trees with up to 90% accuracy.
The research was first published in GECCO 2017 in Berlin.