Some of the earliest approaches to classification are also among the simplest: they do not induce
concept representations (decision trees neural networks etc.) but exploit simple structures of the learning set such as neighborhoods to classify instances.