1. the complete title of one (or more) paper(s) published in the open literature describing the work that the author claims describes a human-competitive result; A Hybrid Method for Feature Construction and Selection to Improve Wind-Damage Prediction in the Forestry Sector 2. the name, complete physical mailing address, e-mail address, and phone number of EACH author of EACH paper(s); Emma Hart e.hart@napier.ac.uk +44 131 455 2783 Edinburgh Napier University, 10 Colinton Road, Edinburgh, EH10 5DT, Scotland, UK. Kevin Sim k.sim@napier.ac.uk +44 131 455 2783 Edinburgh Napier University 10 Colinton Road, Edinburgh, EH10 5DT, Scotland, UK. Barry Gardiner barry.gardiner@inra.fr Ê+33-5-57122855 ISPA, INRA, Bordeaux Sciences, Agro, Villenave dÕOrno, France Kana Kamimura kamimura@shinshu-u.ac.jp Institute of Mountain Science, Shinshu University, 8304, Minamiminowa-Village, Kamiina-County 399-4598, Japan. 3. the name of the corresponding author (i.e., the author to whom notices will be sent concerning the competition); Emma Hart 4. the abstract of the paper(s); Catastrophic damage to forests resulting from major storms has resulted in serious timber and financial losses within the sector across Europe in the recent past. Developing risk assessment methods is thus one of the keys to finding forest management strategies to reduce future damage. Previous approaches to predicting damage to individual trees have used mechanistic models of wind-flow or logistical regression with mixed results. We propose a novel filter-based Genetic Programming method for constructing a large set of new features which are ranked using the Hellinger distance metric which is insensitive to skew in the data. A wrapper-based feature-selection method that uses a random forest classifier is then applied predict damage to individual trees. Using data collected from two forests within South-West France, we demonstrate significantly improved classification results using the new features, and in comparison to previously published results. The feature-selection method retains a small set of relevant variables consisting only of newly constructed features whose components provide insights that can inform forest management policies. 5. a list containing one or more of the eight letters (A, B, C, D, E, F, G, or H) that correspond to the criteria (see above) that the author claims that the work satisfies; B. D. E. 6. a statement stating why the result satisfies the criteria that the contestant claims (see examples of statements of human-competitiveness as a guide to aid in constructing this part of the submission); (B) The result is equal to or better than a result that was accepted as a new scientific result at the time when it was published in a peer-reviewed scientific journal: Our approach (which utilises GP to evolve large sets of new features which are then used in a Random-Forest classifier in conjunction with a feature selection technique) significantly improves on the state-of-the-art in wind-damage classification of trees reported in the 2016 paper cited below: the results not only improve the classification accuracy, but also the associated AUC (area-under-the receiver-operating- curve). The latter value is of particular importance to practitioners in the Forestry industry in terms of having confidence in applying the model as part of an enhanced forest management process. (Our approach.) K. Kamimura, B. Gardiner, S. Dupont, D. Guyon, and C. Meredieu. 2016. Mechanistic and statistical approaches to predicting wind damage to individual maritime pine (Pinus pinaster) trees in forests. Canadian Journal of Forest Research 46 (2016), 88Ð100. (D) The result is publishable in its own right as a new scientific resultÊindependent of the fact that the result was mechanically created. The result is publishable in its own right for two reasons; firstly, forestry management are specifically interested in obtaining accurate models that enable them to develop new policies to better manage forests therefore the resulting accurate model is important and of practical use, regardless of how the result was generated. Secondly, forestry researchers have scientific interest in gaining insight into the factors that influence damage; the evolved features highlight relationships between original variables and those that are frequently used and the results add new insight here. (E): The result is equal to or better than the most recent human-created solution to a long-standing problem for which there has been a succession of increasingly better human-created solutions. Understanding the mechanisms that lead to wind-damage in forest and being able predict them is a long-standing problem in the forestry industry. Previous approaches to classification of wind-damage in forest have used either airflow models coupled with mechanistic wind-prediction models or statistical methods (biased logistic regression) Ð a line of research in this respect goes back to around 2000. Our novel GP based approach that evolves new features outperforms both the mechanistic and statistical methods, as well as state-of-the-art ensemble classifiers (e.g. Random Forests) when applied using only the original features. 7. a full citation of the paper (that is, author names; publication date; name of journal, conference, technical report, thesis, book, or book chapter; name of editors, if applicable, of the journal or edited book; publisher name; publisher city; page numbers, if applicable); Emma Hart, Kevin Sim, Barry Gardiner, and Kana Kamimura. 2017. A hybrid method for feature construction and selection to improve wind-damage prediction in the forestry sector. InÊProceedings of the Genetic and Evolutionary Computation ConferenceÊ(GECCO '17). ACM, New York, NY, USA, 1121-1128. DOI: https://doi.org/10.1145/3071178.3071217 8. a statement either that "any prize money, if any, is to be divided equally among the co-authors" OR a specific percentage breakdown as to how the prize money, if any, is to be divided among the co-authors. Any prize money, if any, is to be divided equally among the co-authors 9. a statement stating why the authors expect that their entry would be the "best," Strong winds during storms can cause catastrophic damage to forests: in 1999, a storm in South Western France caused 26 million m3 of timber loss, which was equivalent to the general harvested volume in the region for 3.5 years; in 2009, a storm in the same area led to losses of approximately 1.8 billion Euros in the forestry sector, almost 60% of total economic losses in France that year. As storms are predicted to become more intense due to climate change, it is crucial to understand the direct causes leading to damage occurrence and develop methodologies to assess and predict the risk of damage in order to sustainably manage the forests. Our approach obtains significantly better results that previous mechanistic or statistical processes: 79% (c.f. 63%) in one forest and 90% in another (c.f. 72%), with AUC values 0.85 and 0.94 from the GP compared to values of 0.71 and 0.76 from previous work. It should be considered the best as it both provides a new scientific method and a state-of-the-art result, and because application of the result has potential to achieve considerable economic impact through informing forestry management policy development. Finally, a recent article* describing this work published in The Conversation has had ~3000 reads since April 23rd 2018, raising the profile of practical Evolutionary Computing. *Storm damage to forests costs billions Ð here's how artificial intelligence can help. The Conversation, April 23rd 2018, Emma Hart and Barry Gardiner. https://bit.ly/2smvN2x 10. An indication of the general type of evolutionary computing used: Genetic Programming 11. The date of publication of the paper July 2017