How to Detect Topology of a Manifold to Approximate Multidimensional Data : доклад, тезисы доклада

Описание

Тип публикации: доклад, тезисы доклада, статья из сборника материалов конференций

Конференция: Applied Methods of Statistical Analysis. Nonparametric Methods in Cybernetics and System Analysis (AMSA'2017); Krasnoyarsk; Krasnoyarsk

Год издания: 2017

Ключевые слова: order, unexpectedness, cluster, connectivity

Аннотация: New method is proposed to identify topology of a low-dimensional manifold approximating multidimensional datasets. The method is based on the implementation of the compliment for the discrete set of data. Some essential properties and constraints of the method are discussed. New method is proposed to identify clusters in datasets. Показать полностьюThe method is based on a sequential elimination of the longest distances in dataset, so that the relevant graph looses some edges. The method stops when the graph becomes disconnected.

Ссылки на полный текст

Издание

Журнал: Applied Methods of Statistical Analysis. Nonparametric Methods in Cybernetics and System Analysis (AMSA'2017)

Номера страниц: 204-210

Издатель: Новосибирский государственный технический университет

Персоны

Вхождение в базы данных