. . . . . "Meta-Learning for Hierarchical Classification and Applications in Bioinformatics"^^ . . . "12" . . . . . . . . "2018-07-01" . . . . . . . . . . "7" . . . . . "Hierarchical classification is a special type of\r\nclassification task where the class labels are organised into a\r\nhierarchy, with more generic class labels being ancestors of more\r\nspecific ones. Meta-learning for classification-algorithm\r\nrecommendation consists of recommending to the user a classification\r\nalgorithm, from a pool of candidate algorithms, for a dataset, based on\r\nthe past performance of the candidate algorithms in other datasets.\r\nMeta-learning is normally used in conventional, non-hierarchical\r\nclassification. By contrast, this paper proposes a meta-learning\r\napproach for more challenging task of hierarchical classification, and\r\nevaluates it in a large number of bioinformatics datasets. Hierarchical\r\nclassification is especially relevant for bioinformatics problems, as\r\nprotein and gene functions tend to be organised into a hierarchy of\r\nclass labels.\r\nThis work proposes meta-learning approach for\r\nrecommending the best hierarchical classification algorithm to a\r\nhierarchical classification dataset. This work\u2019s contributions are: 1)\r\nproposing an algorithm for splitting hierarchical datasets into\r\nnew datasets to increase the number of meta-instances, 2) proposing\r\nmeta-features for hierarchical classification, and 3) interpreting\r\ndecision-tree meta-models for hierarchical classification algorithm\r\nrecommendation."^^ . . . . . . .