The application of machine learning (ML) to solve practicalproblems is complex. Only recently, due to the increased promise ofML in solving real problems and the experienced difficulty of theiruse, has this issue started to attract attention. This difficultyarises from the complexity of learning problems and the large varietyof available techniques. In order to understand this complexity andbegin to overcome it, it is important to construct a characterizationof learning situations. Building on previous work that dealt with thepractical use of ML, a set of dimensions is developed, contrated withanother recent proposal, and illustrated with a project on thedevelopment of a decision-support system for marine propeller design.The general research opportunities that emerge from the developmentof the dimensions are discussed. Leading toward working systems, asimple model is presented for setting priorities in research and inselecting learning tasks within large projects. Central to thedevelopment of the concepts discussed in this paper is their use infuture projects and the recording of their successes, limitations,and failures.
展开▼
